@misc{shen2026embodiedswe,
  title        = {{EmbodiedSWE: Coding Agents for Long Horizon Dexterous Robotics}},
  author       = {Shen, Zeyu and You, Haoxiang and Liu, Yilang and Zheng, Zhicheng and Zha, Lihan and Yamazaki, Kashu and Zhang, Mingtong and Huang, Suning and Sun, Jiankai and Chen, Qianzhong and He, Lucy and Liu, Kaiyuan and Chang, Haoran and Fragkiadaki, Katerina and Shah, Dhruv and Schwager, Mac and Henderson, Peter and Abraham, Ian and Xu, Canwen},
  year         = {2026},
  eprint       = {2609.27308},
  archivePrefix = {arXiv},
  primaryClass  = {cs.RO},
  doi          = {10.48550/arXiv.2609.27308},
  url          = {https://arxiv.org/abs/2609.27308}
}

% references.bib — the single bibliography for this manuscript.
% Formerly references_v6.bib; the legacy SkillOpt-Lite bib (different cite-key
% convention, no key overlap) was deleted on 2026-09-07 and is recoverable from
% git history if skillopt-lite/ prose is ever transcribed.
% Active literature-review entries use the same stable citation IDs; unresolved placeholder entries were removed on 2026-09-09.
%
% RE-AUDIT 2026-09-10. Every entry was re-verified against primary sources:
%   arXiv API (all 121 eprint IDs), Crossref (all DOIs + a title sweep of every
%   arXiv-only entry), the official proceedings indexes of ICLR 2023-2026,
%   NeurIPS 2022-2025, ICML 2024-2026 (PMLR v202/v235/v267 + icml.cc/virtual/2026),
%   COLM 2025-2026, ACL Anthology, JMLR, and the TMLR paper list on jmlr.org.
%   Where a venue is claimed, it was confirmed on that venue's own site.
%   DBLP and OpenReview were unreachable (bot challenge), so no claim in this file
%   rests on them; entries that could not be confirmed at any venue stay as arXiv
%   preprints rather than being upgraded on secondary evidence.
%   Shorthand used below: "no venue found (2026-09 re-audit)" means the title was
%   searched in every source listed above with no hit.

@misc{wang2024benchmarkselfevolving,
  title         = {Benchmark Self-Evolving: A Multi-Agent Framework for Dynamic {LLM} Evaluation},
  author        = {Wang, Siyuan and Long, Zhuohan and Fan, Zhihao and Wei, Zhongyu and Huang, Xuanjing},
  year          = {2024},
  eprint        = {2402.11443},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2402.11443}
}

@misc{white2025livebench,
  title         = {{LiveBench}: A Challenging, Contamination-Limited {LLM} Benchmark},
  author        = {White, Colin and Dooley, Samuel and Roberts, Manley and Pal, Arka and Feuer, Ben and Jain, Siddhartha and Shwartz-Ziv, Ravid and Jain, Neel and Saifullah, Khalid and Dey, Sreemanti and Shubh-Agrawal and Sandha, Sandeep Singh and Naidu, Siddartha and Hegde, Chinmay and LeCun, Yann and Goldstein, Tom and Neiswanger, Willie and Goldblum, Micah},
  year          = {2024},
  eprint        = {2406.19314},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  note          = {ICLR 2025 Spotlight},
  url           = {https://arxiv.org/abs/2406.19314}
}

@misc{yi2025upbench,
  title         = {{UpBench}: A Dynamically Evolving Real-World Labor-Market Agentic Benchmark Framework Built for Human-Centric {AI}},
  author        = {Yi, Darvin and Liu, Teng and Terzolo, Mattie and Hasson, Lance and Sinha, Ayan and Mendes, Pablo N. and Rabinovich, Andrew},
  year          = {2025},
  eprint        = {2511.12306},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2511.12306}
}

@inproceedings{chen2022memory,
  title     = {Memory Bounds for Continual Learning},
  author    = {Chen, Xi and Papadimitriou, Christos and Peng, Binghui},
  booktitle = {2022 IEEE 63rd Annual Symposium on Foundations
               of Computer Science (FOCS)},
  year      = {2022},
  doi       = {10.1109/FOCS54457.2022.00056}
}

@article{shalev2010learnability,
  title   = {Learnability, Stability and Uniform Convergence},
  author  = {Shalev-Shwartz, Shai and Shamir, Ohad
             and Srebro, Nathan and Sridharan, Karthik},
  journal = {Journal of Machine Learning Research},
  volume  = {11},
  pages   = {2635--2670},
  year    = {2010}
}

@article{hoeffding1963probability,
  title   = {Probability Inequalities for Sums of Bounded
             Random Variables},
  author  = {Hoeffding, Wassily},
  journal = {Journal of the American Statistical Association},
  volume  = {58},
  number  = {301},
  pages   = {13--30},
  year    = {1963}
}

@article{dwork2015holdout,
  title   = {Generalization in Adaptive Data Analysis
             and Holdout Reuse},
  author  = {Dwork, Cynthia and Feldman, Vitaly and Hardt, Moritz
             and Pitassi, Toniann and Reingold, Omer and Roth, Aaron},
  journal = {arXiv preprint arXiv:1506.02629},
  year    = {2015}
}

@misc{yan2026livemedbench,
  title         = {{LiveMedBench}: A Contamination-Free Medical Benchmark for {LLMs} with Automated Rubric Evaluation},
  author        = {Yan, Zhiling and Song, Dingjie and Fang, Zhe and Ji, Yisheng and Li, Xiang and Li, Quanzheng and Sun, Lichao},
  year          = {2026},
  eprint        = {2602.10367},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2602.10367}
}

@misc{xu2026mathduels,
  title         = {{MathDuels}: A Self-Play Benchmark That Grows},
  author        = {Xu, Zhiqiu and Jin, Shibo and Arya, Shreya and Naik, Mayur},
  year          = {2026},
  eprint        = {2604.21916},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  note          = {COLM 2026},
  url           = {https://arxiv.org/abs/2604.21916}
}

@misc{chen2026stresseval,
  title         = {{StressEval}: Failure-Driven Dynamic Benchmarking for Knowledge-Intensive Reasoning in Large Language Models},
  author        = {Chen, Yongrui and Ma, Yangyang and Huang, Xiaoying and Zhang, Shenyu and Chen, Huajun and Wang, Haofen and Qi, Guilin},
  year          = {2026},
  eprint        = {2605.01939},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2605.01939}
}

% OLDKEY: arxiv2303-11366 -> NEWKEY: shinn2023reflexion
@inproceedings{shinn2023reflexion,
  title     = {Reflexion: Language Agents with Verbal Reinforcement Learning},
  author    = {Shinn, Noah and Cassano, Federico and Gopinath, Ashwin and Narasimhan, Karthik and Yao, Shunyu},
  booktitle = {Advances in Neural Information Processing Systems},
  volume    = {36},
  year      = {2023},
  url       = {http://papers.nips.cc/paper_files/paper/2023/hash/1b44b878bb782e6954cd888628510e90-Abstract-Conference.html}
}

% OLDKEY: arxiv2305-16291 -> NEWKEY: wang2024voyager
@article{wang2024voyager,
  title   = {Voyager: An Open-Ended Embodied Agent with Large Language Models},
  author  = {Wang, Guanzhi and Xie, Yuqi and Jiang, Yunfan and Mandlekar, Ajay and Xiao, Chaowei and Zhu, Yuke and Fan, Linxi and Anandkumar, Anima},
  journal = {Transactions on Machine Learning Research},
  volume  = {2024},
  year    = {2024},
  url     = {https://openreview.net/forum?id=ehfRiF0R3a}
}

% OLDKEY: arxiv2308-10144 -> NEWKEY: zhao2024expel
@inproceedings{zhao2024expel,
  title     = {{ExpeL}: {LLM} Agents Are Experiential Learners},
  author    = {Zhao, Andrew and Huang, Daniel and Xu, Quentin and Lin, Matthieu and Liu, Yong-Jin and Huang, Gao},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  volume    = {38},
  pages     = {19632--19642},
  publisher = {AAAI Press},
  year      = {2024},
  doi       = {10.1609/aaai.v38i17.29936}
}

% OLDKEY: arxiv2504-07952 -> NEWKEY: suzgun2026dynamic
@inproceedings{suzgun2026dynamic,
  title     = {Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory},
  author    = {Suzgun, Mirac and Yuksekgonul, Mert and Bianchi, Federico and Jurafsky, Dan and Zou, James},
  booktitle = {Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)},
  month     = mar,
  year      = {2026},
  address   = {Rabat, Morocco},
  publisher = {Association for Computational Linguistics},
  pages     = {7080--7106},
  doi       = {10.18653/v1/2026.eacl-long.333},
  url       = {https://aclanthology.org/2026.eacl-long.333/}
}

% OLDKEY: arxiv2509-25140 -> NEWKEY: ouyang2026reasoningbank
@inproceedings{ouyang2026reasoningbank,
  title     = {{ReasoningBank}: Scaling Agent Self-Evolving with Reasoning Memory},
  author    = {Ouyang, Siru and Yan, Jun and Hsu, I-Hung and Chen, Yanfei and Jiang, Ke and Wang, Zifeng and Han, Rujun and Le, Long T. and Daruki, Samira and Tang, Xiangru and Tirumalashetty, Vishy and Lee, George and Rofouei, Mahsan and Lin, Hangfei and Han, Jiawei and Lee, Chen-Yu and Pfister, Tomas},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://openreview.net/forum?id=jL7fwchScm}
}

% OLDKEY: arxiv2508-06433 -> NEWKEY: fang2026memp
@inproceedings{fang2026memp,
  title     = {Memp: Exploring Agent Procedural Memory},
  author    = {Fang, Runnan and Liang, Yuan and Wang, Xiaobin and Wu, Jialong and Qiao, Shuofei and Xie, Pengjun and Huang, Fei and Chen, Huajun and Zhang, Ningyu},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  month     = jul,
  year      = {2026},
  address   = {San Diego, California, United States},
  publisher = {Association for Computational Linguistics},
  pages     = {17490--17502},
  doi       = {10.18653/v1/2026.findings-acl.866}
}

% OLDKEY: arxiv2510-04618 -> NEWKEY: zhang2026agentic
@inproceedings{zhang2026agentic,
  title     = {Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models},
  author    = {Zhang, Qizheng and Hu, Changran and Upasani, Shubhangi and Ma, Boyuan and Hong, Fenglu and Kamanuru, Vamsidhar and Rainton, Jay and Wu, Chen and Ji, Mengmeng and Li, Hanchen and Thakker, Urmish and Zou, James and Olukotun, Kunle},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://openreview.net/forum?id=eC4ygDs02R}
}

% OLDKEY: arxiv2409-07429 -> NEWKEY: wang2025workflow
@inproceedings{wang2025workflow,
  title        = {Agent Workflow Memory},
  author       = {Wang, Zora Zhiruo and Mao, Jiayuan and Fried, Daniel and Neubig, Graham},
  booktitle    = {Proceedings of the 42nd International Conference on Machine Learning},
  series       = {Proceedings of Machine Learning Research},
  volume       = {267},
  pages        = {63897--63911},
  year         = {2025},
  publisher    = {PMLR},
}

% OLDKEY: arxiv2510-12635 -> NEWKEY: zhang2026action
% Published version located in the 2026-09-10 re-audit (Crossref); was previously listed as arXiv-only.
@inproceedings{zhang2026action,
  title     = {Memory as Action: Autonomous Context Curation for Long-Horizon Agentic Tasks},
  author    = {Zhang, Yuxiang and Shu, Jiangming and Ma, Ye and Lin, Xueyuan and Wu, Shangxi and Sang, Jitao},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  month     = jul,
  year      = {2026},
  address   = {San Diego, California, United States},
  publisher = {Association for Computational Linguistics},
  pages     = {19149--19164},
  doi       = {10.18653/v1/2026.findings-acl.956},
  url       = {https://aclanthology.org/2026.findings-acl.956/}
}

% OLDKEY: arxiv2605-09998 -> NEWKEY: karten2026harness
@misc{karten2026harness,
  title         = {Continual Harness: Online Adaptation for Self-Improving Foundation Agents},
  author        = {Karten, Seth and Zhang, Joel and Upaa, Jr., Tersoo and Feng, Ruirong and Li, Wenzhe and Shi, Chengshuai and Jin, Chi and Vodrahalli, Kiran},
  year          = {2026},
  eprint        = {2605.09998},
  archivePrefix = {arXiv},
  note          = {arXiv:2605.09998}
}

% OLDKEY: arxiv2603-25158 -> NEWKEY: ni2026trace2skill
@misc{ni2026trace2skill,
  title         = {{Trace2Skill}: Distill Trajectory-Local Lessons into Transferable Agent Skills},
  author        = {Ni, Jingwei and Liu, Yihao and Liu, Xinpeng and Sun, Yutao and Zhou, Mengyu and Cheng, Pengyu and Wang, Dexin and Zhao, Erchao and Jiang, Xiaoxi and Jiang, Guanjun},
  year          = {2026},
  eprint        = {2603.25158},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {arXiv:2603.25158}
}

% OLDKEY: arxiv2211-01910 -> NEWKEY: zhou2023large
@inproceedings{zhou2023large,
  title     = {Large Language Models Are Human-Level Prompt Engineers},
  author    = {Zhou, Yongchao and Muresanu, Andrei Ioan and Han, Ziwen and Paster, Keiran and Pitis, Silviu and Chan, Harris and Ba, Jimmy},
  booktitle = {The Eleventh International Conference on Learning Representations},
  year      = {2023},
  url       = {https://openreview.net/forum?id=92gvk82DE-}
}

% OLDKEY: arxiv2309-03409 -> NEWKEY: yang2024large
@inproceedings{yang2024large,
  title     = {Large Language Models as Optimizers},
  author    = {Yang, Chengrun and Wang, Xuezhi and Lu, Yifeng and Liu, Hanxiao and Le, Quoc V. and Zhou, Denny and Chen, Xinyun},
  booktitle = {The Twelfth International Conference on Learning Representations},
  year      = {2024},
  url       = {https://openreview.net/forum?id=Bb4VGOWELI}
}

% OLDKEY: arxiv2309-16797 -> NEWKEY: fernando2024promptbreeder
@inproceedings{fernando2024promptbreeder,
  title        = {Promptbreeder: Self-Referential Self-Improvement via Prompt Evolution},
  author       = {Fernando, Chrisantha and Banarse, Dylan Sunil and Michalewski, Henryk and Osindero, Simon and Rockt{\"a}schel, Tim},
  booktitle    = {Proceedings of the 41st International Conference on Machine Learning},
  series       = {Proceedings of Machine Learning Research},
  volume       = {235},
  pages        = {13481--13544},
  year         = {2024},
  publisher    = {PMLR},
  url          = {https://proceedings.mlr.press/v235/fernando24a.html}
}

% OLDKEY: arxiv2406-11695 -> NEWKEY: opsahlong2024optimizing
@inproceedings{opsahlong2024optimizing,
  title     = {Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programs},
  author    = {Opsahl-Ong, Krista and Ryan, Michael J. and Purtell, Josh and Broman, David and Potts, Christopher and Zaharia, Matei and Khattab, Omar},
  booktitle = {Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing},
  month     = nov,
  year      = {2024},
  address   = {Miami, Florida, USA},
  publisher = {Association for Computational Linguistics},
  pages     = {9340--9366},
  doi       = {10.18653/v1/2024.emnlp-main.525},
  url       = {https://aclanthology.org/2024.emnlp-main.525/}
}

% OLDKEY: arxiv2408-08435 -> NEWKEY: hu2025automated
@inproceedings{hu2025automated,
  title     = {Automated Design of Agentic Systems},
  author    = {Hu, Shengran and Lu, Cong and Clune, Jeff},
  booktitle = {The Thirteenth International Conference on Learning Representations},
  year      = {2025},
  address   = {Singapore},
  url       = {https://openreview.net/forum?id=t9U3LW7JVX}
}

% OLDKEY: arxiv2410-10762 -> NEWKEY: zhang2025aflow
@inproceedings{zhang2025aflow,
  title     = {{AFlow}: Automating Agentic Workflow Generation},
  author    = {Zhang, Jiayi and Xiang, Jinyu and Yu, Zhaoyang and Teng, Fengwei and Chen, Xiong-Hui and Chen, Jiaqi and Zhuge, Mingchen and Cheng, Xin and Hong, Sirui and Wang, Jinlin and Zheng, Bingnan and Liu, Bang and Luo, Yuyu and Wu, Chenglin},
  booktitle = {The Thirteenth International Conference on Learning Representations},
  year      = {2025},
  address   = {Singapore},
  url       = {https://openreview.net/forum?id=z5uVAKwmjf}
}

% OLDKEY: arxiv2502-04180 -> NEWKEY: zhang2025supernet
@inproceedings{zhang2025supernet,
  title        = {Multi-agent Architecture Search via Agentic Supernet},
  author       = {Zhang, Guibin and Niu, Luyang and Fang, Junfeng and Wang, Kun and Bai, Lei and Wang, Xiang},
  booktitle    = {Proceedings of the 42nd International Conference on Machine Learning},
  series       = {Proceedings of Machine Learning Research},
  volume       = {267},
  pages        = {75834--75852},
  year         = {2025},
  publisher    = {PMLR},
  url          = {https://proceedings.mlr.press/v267/zhang25bi.html}
}

% OLDKEY: arxiv2410-06153 -> NEWKEY: shang2025agentsquare
@inproceedings{shang2025agentsquare,
  title     = {{AgentSquare}: Automatic {LLM} Agent Search in Modular Design Space},
  author    = {Shang, Yu and Li, Yu and Zhao, Keyu and Ma, Likai and Liu, Jiahe and Xu, Fengli and Li, Yong},
  booktitle = {The Thirteenth International Conference on Learning Representations},
  year      = {2025},
  address   = {Singapore},
  url       = {https://openreview.net/forum?id=mPdmDYIQ7f}
}

% OLDKEY: arxiv2206-08896 -> NEWKEY: lehman2023evolution
@incollection{lehman2023evolution,
  author    = {Joel Lehman and Jonathan Gordon and Shawn Jain and Kamal Ndousse and Cathy Yeh and Kenneth O. Stanley},
  title     = {Evolution Through Large Models},
  booktitle = {Handbook of Evolutionary Machine Learning},
  editor    = {Wolfgang Banzhaf and Penousal Machado and Mengjie Zhang},
  series    = {Genetic and Evolutionary Computation},
  pages     = {331--366},
  publisher = {Springer},
  address   = {Singapore},
  year      = {2023},
  doi       = {10.1007/978-981-99-3814-8_11}
}

% OLDKEY: arxiv2511-23473 -> NEWKEY: wang2026thetaevolve
% Accepted at ICML 2026; confirmed on icml.cc/virtual/2026 (2026-09-10 re-audit).
% PMLR volume/pages not yet assigned, so they are omitted rather than invented.
@inproceedings{wang2026thetaevolve,
  title     = {ThetaEvolve: Test-time Learning on Open Problems},
  author    = {Wang, Yiping and Su, Shao-Rong and Zeng, Zhiyuan and Xu, Eva and Ren, Liliang and Yang, Xinyu and Huang, Zeyi and He, Xuehai and Ma, Luyao and Peng, Baolin and Cheng, Hao and He, Pengcheng and Chen, Weizhu and Wang, Shuohang and Du, Simon Shaolei and Shen, Yelong},
  booktitle = {Proceedings of the 43rd International Conference on Machine Learning (ICML 2026)},
  year      = {2026},
  url       = {https://icml.cc/virtual/2026/poster/63428},
  note      = {arXiv:2511.23473}
}

% OLDKEY: arxiv2506-13131 -> NEWKEY: novikov2025alphaevolve
@misc{novikov2025alphaevolve,
  author        = {Alexander Novikov and Ng{\^a}n V{\~u} and Marvin Eisenberger and Emilien Dupont and Po-Sen Huang and Adam Zsolt Wagner and Sergey Shirobokov and Borislav Kozlovskii and Francisco J. R. Ruiz and Abbas Mehrabian and M. Pawan Kumar and Abigail See and Swarat Chaudhuri and George Holland and Alex Davies and Sebastian Nowozin and Pushmeet Kohli and Matej Balog},
  title         = {{AlphaEvolve}: A Coding Agent for Scientific and Algorithmic Discovery},
  year          = {2025},
  eprint        = {2506.13131},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {Google DeepMind technical report},
  url           = {https://arxiv.org/abs/2506.13131}
}

% OLDKEY: arxiv2509-19349 -> NEWKEY: lange2026shinkaevolve
% ICLR 2026 acceptance confirmed in the official proceedings index (2026-09-10 re-audit).
@inproceedings{lange2026shinkaevolve,
  title     = {ShinkaEvolve: Towards Open-Ended and Sample-Efficient Program Evolution},
  author    = {Lange, Robert Tjarko and Imajuku, Yuki and Cetin, Edoardo},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2026/hash/7886b9bafe76c52fd568db10ff9772df-Abstract-Conference.html},
  note      = {Poster; originally released as arXiv:2509.19349}
}

% OLDKEY: arxiv2505-22954 -> NEWKEY: zhang2026darwin
@inproceedings{zhang2026darwin,
  title     = {Darwin {G{\"o}del} Machine: Open-Ended Evolution of Self-Improving Agents},
  author    = {Zhang, Jenny and Hu, Shengran and Lu, Cong and Lange, Robert Tjarko and Clune, Jeff},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  note      = {Poster},
  url       = {https://openreview.net/forum?id=pUpzQZTvGY}
}

% OLDKEY: arxiv2504-15228 -> NEWKEY: robeyns2025selfimproving
% No peer-reviewed venue found (2026-09 re-audit). The arXiv "Comments" field reads
% "Submitted as a preprint to NeurIPS 2025" -- i.e. no acceptance is claimed by the authors.
@misc{robeyns2025selfimproving,
  title         = {A Self-Improving Coding Agent},
  author        = {Robeyns, Maxime and Szummer, Martin and Aitchison, Laurence},
  year          = {2025},
  eprint        = {2504.15228},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {arXiv preprint},
  url           = {https://arxiv.org/abs/2504.15228}
}

% OLDKEY: arxiv2410-04444 -> NEWKEY: yin2025godel
% NOTE: "Recursively Self-Improvement" is the official ACL Anthology / Crossref title (sic);
% the arXiv preprint reads "Recursive Self-Improvement". The published form is used here.
@inproceedings{yin2025godel,
  title     = {{G{\"o}del} Agent: A Self-Referential Agent Framework for Recursively Self-Improvement},
  author    = {Yin, Xunjian and Wang, Xinyi and Pan, Liangming and Lin, Li and Wan, Xiaojun and Wang, William Yang},
  booktitle = {Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  pages     = {27890--27913},
  year      = {2025},
  month     = jul,
  address   = {Vienna, Austria},
  publisher = {Association for Computational Linguistics},
  doi       = {10.18653/v1/2025.acl-long.1354},
  url       = {https://aclanthology.org/2025.acl-long.1354/}
}

% OLDKEY: arxiv2507-19457 -> NEWKEY: agrawal2026gepa
@inproceedings{agrawal2026gepa,
  title     = {{GEPA}: Reflective Prompt Evolution Can Outperform Reinforcement Learning},
  author    = {Agrawal, Lakshya A and Tan, Shangyin and Soylu, Dilara and Ziems, Noah and Khare, Rishi and Opsahl-Ong, Krista and Singhvi, Arnav and Shandilya, Herumb and Ryan, Michael J and Jiang, Meng and Potts, Christopher and Sen, Koushik and Dimakis, Alexandros G. and Stoica, Ion and Klein, Dan and Zaharia, Matei and Khattab, Omar},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  note      = {Oral},
  url       = {https://openreview.net/forum?id=RQm2KQTM5r}
}

% OLDKEY: arxiv2605-24539 -> NEWKEY: che2026demoevolve
@misc{che2026demoevolve,
  title         = {DemoEvolve: Overcoming Sparse Feedback in Agentic Harness Evolution with Demonstrations},
  author        = {Che, Lirong and Yang, Yuzhe and Lin, Peiwen and Wang, Chuang and Wang, Xueqian and Su, Jian},
  year          = {2026},
  eprint        = {2605.24539},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {arXiv preprint},
  url           = {https://arxiv.org/abs/2605.24539}
}

% OLDKEY: arxiv2606-13317 -> NEWKEY: chen2026skillcat
@misc{chen2026skillcat,
  title         = {SkillCAT: Contrastive, Assessment-Augmented and Topology-Aware Skill Self-Evolution for LLM Agents},
  author        = {Chen, Kunfeng and Zhong, Qihuang and Liu, Juhua and Du, Bo},
  year          = {2026},
  eprint        = {2606.13317},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  note          = {Preprint, arXiv:2606.13317}
}

% OLDKEY: arxiv2606-01311 -> NEWKEY: yu2026skilladaptor
@misc{yu2026skilladaptor,
  title         = {SkillAdaptor: Self-Adapting Skills for LLM Agents from Trajectories},
  author        = {Yu, Zhuoyun and Xie, Xin and Yao, Wuguannan and Wang, Chenxi and Liang, Lei and Qi, Xiang and Deng, Shumin},
  year          = {2026},
  eprint        = {2606.01311},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  note          = {Preprint, arXiv:2606.01311}
}

% OLDKEY: arxiv2605-23904 -> NEWKEY: yang2026skillopt
@misc{yang2026skillopt,
  title         = {{SkillOpt}: Executive Strategy for Self-Evolving Agent Skills},
  author        = {Yang, Yifan and Gong, Ziyang and Huang, Weiquan and Yang, Qihao and Zhou, Ziwei and Huang, Zisu and Li, Yan and Gao, Xuemei and Dai, Qi and Liu, Bei and Qiu, Kai and Yang, Yuqing and Chen, Dongdong and Yang, Xue and Luo, Chong},
  year          = {2026},
  eprint        = {2605.23904},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {Preprint, arXiv:2605.23904}
}

% OLDKEY: arxiv2607-03451 -> NEWKEY: shen2026skilloptlite
@misc{shen2026skilloptlite,
  title         = {{SkillOpt-Lite}: Better and Faster Agent Self-evolution via One Line of Vibe},
  author        = {Shen, Yifei and Li, Bo and Zhang, Xinjie},
  year          = {2026},
  eprint        = {2607.03451},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SE},
  note          = {Preprint, arXiv:2607.03451}
}

% OLDKEY: arxiv2606-09498 -> NEWKEY: zhang2026selfharness
@misc{zhang2026selfharness,
  title         = {Self-Harness: Harnesses That Improve Themselves},
  author        = {Zhang, Hangfan and Zhang, Shao and Li, Kangcong and Zhang, Chen and Chen, Yang and Zhang, Yiqun and Bai, Lei and Hu, Shuyue},
  year          = {2026},
  eprint        = {2606.09498},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  note          = {Preprint, arXiv:2606.09498}
}

% OLDKEY: arxiv2605-18747 -> NEWKEY: ning2026harness
% NOTE: the arXiv-registered title is "Code as Agent Harness"; the phrase "Toward Executable, Verifiable, and Stateful Agent Systems" comes from the abstract, not the official title.
@misc{ning2026harness,
  title         = {Code as Agent Harness},
  author        = {Ning, Xuying and Tieu, Katherine and Fu, Dongqi and Wei, Tianxin and Li, Zihao and Bei, Yuanchen and Zou, Jiaru and Ai, Mengting and Liu, Zhining and Li, Ting-Wei and Chen, Lingjie and Zhao, Yanjun and Yang, Ke and Li, Bingxuan and Qian, Cheng and Li, Gaotang and Lin, Xiao and Zeng, Zhichen and Qiu, Ruizhong and Chen, Sirui and Sun, Yifan and Yang, Xiyuan and Wang, Ruida and Pan, Rui and Yang, Chenyuan and Zhang, Dylan and Fang, Liri and Cui, Zikun and Cao, Yang and Chen, Pan and Sun, Dorothy and Chen, Ren and Srinivasan, Mahesh and Mathur, Nipun and Xia, Yinglong and Li, Hong and Yan, Hong and Lu, Pan and Zhang, Lingming and Zhang, Tong and Tong, Hanghang and He, Jingrui},
  year          = {2026},
  eprint        = {2605.18747},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  note          = {arXiv preprint},
  url           = {https://arxiv.org/abs/2605.18747}
}

% OLDKEY: arxivcs0309048 -> NEWKEY: schmidhuber2003godel
% NOTE: arXiv:cs/0309048 itself was never formally published. Its own journal-ref field states that
% *variants* appeared in "Adaptive Agents and Multi-Agent Systems II" (LNCS 3394, pp. 1-23, Springer,
% 2005), in Proc. ICANN 2005 (LNCS 3697, pp. 223-233), and in "Artificial General Intelligence"
% (Cognitive Technologies, Springer, 2006). Those are variants with different titles, not a
% republication of this preprint, so substituting one would silently change the cited work.
% The literal preprint is kept; swap it only if you deliberately want to cite a specific variant.
@misc{schmidhuber2003godel,
  title = {{G{\"o}del} Machines: Self-Referential Universal Problem Solvers Making Provably Optimal Self-Improvements},
  author = {J{\"u}rgen Schmidhuber},
  year = {2003},
  eprint = {cs/0309048},
  archivePrefix = {arXiv},
  note = {arXiv:cs/0309048}
}

% OLDKEY: weng2026harness -> NEWKEY: weng2026harness
@misc{weng2026harness,
  title        = {Harness Engineering for Self-Improvement},
  author       = {Weng, Lilian},
  year         = {2026},
  howpublished = {\url{https://lilianweng.github.io/posts/2026-07-04-harness/}},
  note         = {Blog post, published 2026-07-04}
}

% OLDKEY: good1965ultraintelligent -> NEWKEY: good1965ultraintelligent
@incollection{good1965ultraintelligent,
  title     = {Speculations Concerning the First Ultraintelligent Machine},
  author    = {Good, Irving John},
  booktitle = {Advances in Computers},
  editor    = {Alt, Franz L. and Rubinoff, Morris},
  volume    = {6},
  pages     = {31--88},
  year      = {1965},
  publisher = {Elsevier}
}

% OLDKEY: shalevshwartz2010learnability -> NEWKEY: shalevshwartz2010learnability
@article{shalevshwartz2010learnability,
  title   = {Learnability, Stability and Uniform Convergence},
  author  = {Shalev-Shwartz, Shai and Shamir, Ohad and Srebro, Nathan and Sridharan, Karthik},
  journal = {Journal of Machine Learning Research},
  volume  = {11},
  pages   = {2635--2670},
  year    = {2010}
}

% OLDKEY: liu2020zoprimer -> NEWKEY: liu2020zoprimer
% NOTE: "Principals" (not "Principles") is the spelling in the official IEEE record and in the
% published article's title; kept verbatim rather than silently corrected.
@article{liu2020zoprimer,
  title   = {A Primer on Zeroth-Order Optimization in Signal Processing and Machine Learning: Principals, Recent Advances, and Applications},
  author  = {Liu, Sijia and Chen, Pin-Yu and Kailkhura, Bhavya and Zhang, Gaoyuan and Hero, Alfred O. and Varshney, Pramod K.},
  journal = {IEEE Signal Processing Magazine},
  volume  = {37},
  number  = {5},
  pages   = {43--54},
  year    = {2020}
}

% OLDKEY: zhang2023mathanalysis -> NEWKEY: zhang2023mathanalysis
@book{zhang2023mathanalysis,
  title     = {Mathematical Analysis of Machine Learning Algorithms},
  author    = {Zhang, Tong},
  publisher = {Cambridge University Press},
  year      = {2023}
}

% OLDKEY: arxiv2608-07545 -> NEWKEY: zhang2026darwinx
@misc{zhang2026darwinx,
  title         = {{DarwinX}: Evolving Agent Harnesses Through Natural Selection},
  author        = {Zhang, Yifan and Dai, Yutong and Tan, Juntao and Yang, Luyu and Mullur, Rishi and Hoang, Thai and Hu, Zhiyuan and Zhu, James and Mui, Phil and Savarese, Silvio and Xu, Ran and Chen, Zeyuan},
  year          = {2026},
  eprint        = {2608.07545},
  archivePrefix = {arXiv},
  primaryClass  = {cs.NE},
  note          = {arXiv preprint},
  url           = {https://arxiv.org/abs/2608.07545}
}

% OLDKEY: arxiv2606-14249 -> NEWKEY: chen2026harnessx
% NOTE: task brief listed the author as "Darwin Agent Team" (organizational), but the arXiv byline (verified via arxiv.org meta tags) lists 14 named individual authors below; "Darwin Agent Team" is only the project's GitHub Pages site name, not an author.
@misc{chen2026harnessx,
  title         = {{HarnessX}: A Composable, Adaptive, and Evolvable Agent Harness Foundry},
  author        = {Chen, Tingyang and Lu, Shuo and Zhao, Kang and Meng, Weicheng and Teng, Hanlin and Li, Tianhao and Li, Chao and Liu, Xule and Liang, Jian and Zhang, Zhizhong and Xie, Yuan and Qu, Heng and Shao, Kun and Luan, Jian},
  year          = {2026},
  eprint        = {2606.14249},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {arXiv preprint},
  url           = {https://arxiv.org/abs/2606.14249}
}

% OLDKEY: arxiv2402-16823 -> NEWKEY: zhuge2024gptswarm
@inproceedings{zhuge2024gptswarm,
  author       = {Mingchen Zhuge and Wenyi Wang and Louis Kirsch and Francesco Faccio and Dmitrii Khizbullin and J{\"u}rgen Schmidhuber},
  title        = {{GPTSwarm}: Language Agents as Optimizable Graphs},
  booktitle    = {Proceedings of the 41st International Conference on Machine Learning},
  series       = {Proceedings of Machine Learning Research},
  volume       = {235},
  pages        = {62743--62767},
  year         = {2024},
  organization = {PMLR}
}

% OLDKEY: arxiv2305-03495 -> NEWKEY: pryzant2023automatic
@inproceedings{pryzant2023automatic,
  title     = {Automatic Prompt Optimization with "Gradient Descent" and Beam Search},
  author    = {Pryzant, Reid and Iter, Dan and Li, Jerry and Lee, Yin Tat and Zhu, Chenguang and Zeng, Michael},
  booktitle = {Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing},
  pages     = {7957--7968},
  year      = {2023},
  address   = {Singapore},
  publisher = {Association for Computational Linguistics},
  doi       = {10.18653/v1/2023.emnlp-main.494},
  url       = {https://aclanthology.org/2023.emnlp-main.494/}
}

% OLDKEY: arxiv2310-02304 -> NEWKEY: zelikman2024stop
@inproceedings{zelikman2024stop,
  title     = {Self-Taught Optimizer ({STOP}): Recursively Self-Improving Code Generation},
  author    = {Zelikman, Eric and Lorch, Eliana and Mackey, Lester and Kalai, Adam Tauman},
  booktitle = {First Conference on Language Modeling},
  year      = {2024},
  note      = {arXiv:2310.02304}
}

% OLDKEY: yudkowsky2008rsi -> NEWKEY: yudkowsky2008artificial
@incollection{yudkowsky2008artificial,
  title     = {Artificial Intelligence as a Positive and Negative Factor in Global Risk},
  author    = {Yudkowsky, Eliezer},
  booktitle = {Global Catastrophic Risks},
  editor    = {Bostrom, Nick and {\'C}irkovi{\'c}, Milan M.},
  pages     = {308--345},
  year      = {2008},
  publisher = {Oxford University Press}
}

% OLDKEY: bousquet2002stability -> NEWKEY: bousquet2002stability
@article{bousquet2002stability,
  title   = {Stability and Generalization},
  author  = {Bousquet, Olivier and Elisseeff, Andr{\'e}},
  journal = {Journal of Machine Learning Research},
  volume  = {2},
  pages   = {499--526},
  year    = {2002}
}

% OLDKEY: arxiv1509-01240 -> NEWKEY: hardt2016train
@inproceedings{hardt2016train,
  title        = {Train Faster, Generalize Better: Stability of Stochastic Gradient Descent},
  author       = {Hardt, Moritz and Recht, Benjamin and Singer, Yoram},
  booktitle    = {International Conference on Machine Learning},
  pages        = {1225--1234},
  organization = {PMLR},
  volume       = {48},
  year         = {2016}
}

% OLDKEY: arxiv1506-02629 -> NEWKEY: dwork2015reusable
% NOTE: this is the Science article, verified via Crossref. It is NOT arXiv:1506.02629 -- that
% eprint is the companion NeurIPS 2015 paper "Generalization in Adaptive Data Analysis and
% Holdout Reuse". The OLDKEY above is a legacy label only.
@article{dwork2015reusable,
  title={The reusable holdout: Preserving validity in adaptive data analysis},
  author={Dwork, Cynthia and Feldman, Vitaly and Hardt, Moritz and Pitassi, Toniann and Reingold, Omer and Roth, Aaron},
  journal={Science},
  volume={349},
  number={6248},
  pages={636--638},
  year={2015},
  publisher={American Association for the Advancement of Science}
}

% OLDKEY: arxiv1502-04585 -> NEWKEY: blum2015ladder
@inproceedings{blum2015ladder,
  title        = {The Ladder: A Reliable Leaderboard for Machine Learning Competitions},
  author       = {Blum, Avrim and Hardt, Moritz},
  booktitle    = {International Conference on Machine Learning},
  pages        = {1006--1014},
  organization = {PMLR},
  volume       = {37},
  year         = {2015}
}

% OLDKEY: cawley2010overfitting -> NEWKEY: cawley2010overfitting
@article{cawley2010overfitting,
  title={On over-fitting in model selection and subsequent selection bias in performance evaluation},
  author={Cawley, Gavin C and Talbot, Nicola LC},
  journal={The Journal of Machine Learning Research},
  volume={11},
  pages={2079--2107},
  year={2010},
  publisher={JMLR. org}
}

@inproceedings{dodge2019show,
  title={Show your work: Improved reporting of experimental results},
  author={Dodge, Jesse and Gururangan, Suchin and Card, Dallas and Schwartz, Roy and Smith, Noah A},
  booktitle={Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)},
  pages={2185--2194},
  year={2019}
}

@article{kirkpatrick2017overcoming,
  title={Overcoming catastrophic forgetting in neural networks},
  author={Kirkpatrick, James and Pascanu, Razvan and Rabinowitz, Neil and Veness, Joel and Desjardins, Guillaume and Rusu, Andrei A and Milan, Kieran and Quan, John and Ramalho, Tiago and Grabska-Barwinska, Agnieszka and others},
  journal={Proceedings of the National Academy of Sciences},
  volume={114},
  number={13},
  pages={3521--3526},
  year={2017},
  publisher={National Academy of Sciences}
}

% ============================================================
% Tier B systems (cited in §4/§5 text, not protocol-audited)
% ============================================================

% OLDKEY: arxiv2406-07496 -> NEWKEY: yuksekgonul2025textgrad
% NOTE: published Nature version adds co-author "Pan Lu", absent from the arXiv author list given in the task.
@article{yuksekgonul2025textgrad,
  title   = {Optimizing generative {AI} by backpropagating language model feedback},
  author  = {Yuksekgonul, Mert and Bianchi, Federico and Boen, Joseph and Liu, Sheng and Lu, Pan and Huang, Zhi and Guestrin, Carlos and Zou, James},
  journal = {Nature},
  volume  = {639},
  number  = {8055},
  pages   = {609--616},
  year    = {2025},
  doi     = {10.1038/s41586-025-08661-4}
}

% OLDKEY: arxiv2604-08618 -> NEWKEY: liu2026skillforge
% Authors, pages and DOI verified via Crossref; booktitle matches the official ACM container title
% (the Industry Track designation is recorded in the note, not in the container title).
@inproceedings{liu2026skillforge,
  title     = {{SkillForge}: Forging Domain-Specific, Self-Evolving Agent Skills in Cloud Technical Support},
  author    = {Liu, Xingyan and Luo, Xiyue and Li, Linyu and Huang, Ganghong and Liu, Jianfeng and Qiao, Honglin},
  booktitle = {Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval},
  pages     = {4763--4768},
  year      = {2026},
  publisher = {ACM},
  doi       = {10.1145/3805712.3808466},
  note      = {Industry Track}
}

% OLDKEY: arxiv2601-21557 -> NEWKEY: ye2026meta
% Acceptance confirmed on icml.cc/virtual/2026 (2026-09-10 re-audit); PMLR volume/pages
% not yet assigned, so they are omitted rather than invented.
@inproceedings{ye2026meta,
  title     = {Meta Context Engineering via Agentic Skill Evolution},
  author    = {Ye, Haoran and He, Xuning and Arak, Vincent and Dong, Haonan and Song, Guojie},
  booktitle = {Proceedings of the 43rd International Conference on Machine Learning (ICML 2026)},
  year      = {2026},
  url       = {https://icml.cc/virtual/2026/poster/64296},
  note      = {arXiv:2601.21557}
}

% OLDKEY: arxiv2603-28052 -> NEWKEY: lee2026metaharness
% Accepted at COLM 2026; confirmed on colm.cc/Conferences/2026/AcceptedPapers, where the author
% list matches this entry exactly (2026-09-10 re-audit). COLM proceedings are not paginated.
@inproceedings{lee2026metaharness,
  title     = {Meta-Harness: End-to-End Optimization of Model Harnesses},
  author    = {Lee, Yoonho and Nair, Roshen S. and Zhang, Qizheng and Lee, Kangwook and Khattab, Omar and Finn, Chelsea},
  booktitle = {Third Conference on Language Modeling (COLM 2026)},
  year      = {2026},
  note      = {Originally released as arXiv:2603.28052}
}

% OLDKEY: arxiv2604-25850 -> NEWKEY: lin2026agentic
% No venue found (2026-09 re-audit); arXiv preprint.
@misc{lin2026agentic,
  title         = {Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnesses},
  author        = {Lin, Jiahang and Liu, Shichun and Pan, Chengjun and Lin, Lizhi and Dou, Shihan and Xi, Zhiheng and Huang, Xuanjing and Yan, Hang and Han, Zhenhua and Gui, Tao and Jiang, Yu-Gang},
  year          = {2026},
  eprint        = {2604.25850},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SE},
  note          = {arXiv:2604.25850}
}

% OLDKEY: arxiv2310-03714 -> NEWKEY: khattab2024dspy
% NOTE: ICLR 2024 camera-ready proceedings (dblp, iclr.cc, proceedings.iclr.cc all agree) list the
% title as "...into State-of-the-Art Pipelines"; arXiv/OpenReview submission and the authors' own
% GitHub citation retain "...into Self-Improving Pipelines" — a genuine title discrepancy between
% preprint and camera-ready. Using the official proceedings title here.
@inproceedings{khattab2024dspy,
  title     = {{DSPy}: Compiling Declarative Language Model Calls into State-of-the-Art Pipelines},
  author    = {Khattab, Omar and Singhvi, Arnav and Maheshwari, Paridhi and Zhang, Zhiyuan and Santhanam, Keshav and Vardhamanan, Sri and Haq, Saiful and Sharma, Ashutosh and Joshi, Thomas T. and Moazam, Hanna and Miller, Heather and Zaharia, Matei and Potts, Christopher},
  booktitle = {The Twelfth International Conference on Learning Representations},
  year      = {2024},
  url       = {https://openreview.net/forum?id=PFS4ffN9Yx}
}

% OLDKEY: arxiv2509-26354 -> NEWKEY: shao2026misevolve
% ICLR 2026 acceptance confirmed in the official proceedings index (2026-09-10 re-audit).
@inproceedings{shao2026misevolve,
  title     = {Your Agent May Misevolve: Emergent Risks in Self-evolving {LLM} Agents},
  author    = {Shao, Shuai and Ren, Qihan and Liu, Dongrui and Qian, Chen and Wei, Boyi and Guo, Dadi and Yang, Jingyi and Song, Xinhao and Zhang, Linfeng and Zhang, Weinan and Shao, Jing},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2026/hash/a24cd16bc361afa78e57d31d34f3d936-Abstract-Conference.html},
  note      = {Author order follows the ICLR proceedings record, which differs from the arXiv byline; originally released as arXiv:2509.26354}
}

% OLDKEY: arxiv2605-30621 -> NEWKEY: lin2026updating
% No venue found (2026-09 re-audit). Full 17-author list restored from arXiv metadata --
% the previous version of this entry listed only the first six authors.
@misc{lin2026updating,
  title         = {Harness Updating Is Not Harness Benefit: Disentangling Evolution Capabilities in Self-Evolving {LLM} Agents},
  author        = {Lin, Minhua and Wu, Juncheng and Wang, Zijun and Shi, Zhan and Sang, Yisi and He, Bing and Liu, Zewen and Wei, Tianxin and Wu, Zongyu and Zhang, Zhiwei and Wang, Dakuo and Zhang, Xiang and Dumoulin, Benoit and Xie, Cihang and Zhou, Yuyin and Wang, Suhang and Lu, Hanqing},
  year          = {2026},
  eprint        = {2605.30621},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {arXiv:2605.30621}
}

% OLDKEY: arxiv2602-20133 -> NEWKEY: cemri2026adaevolve
% Paper text itself states "Preprint. Under review."; no venue found (2026-09 re-audit).
@misc{cemri2026adaevolve,
  title         = {AdaEvolve: Adaptive LLM Driven Zeroth-Order Optimization},
  author        = {Cemri, Mert and Agrawal, Shubham and Gupta, Akshat and Liu, Shu and Cheng, Audrey and Mang, Qiuyang and Naren, Ashwin and Erdogan, Lutfi Eren and Sen, Koushik and Zaharia, Matei and Dimakis, Alex and Stoica, Ion},
  year          = {2026},
  eprint        = {2602.20133},
  archivePrefix = {arXiv},
  primaryClass  = {cs.NE},
  note          = {arXiv:2602.20133}
}

% OLDKEY: arxiv2506-10943 -> NEWKEY: zweiger2025selfadapting
% NOTE: author list follows the official NeurIPS 2025 proceedings record (papers.nips.cc), which
% lists five authors and the form "Jyo Pari". The arXiv version additionally lists Ekin Aky{\"u}rek
% and spells the second author "Jyothish Pari"; the published record takes precedence here.
@inproceedings{zweiger2025selfadapting,
  title     = {Self-Adapting Language Models},
  author    = {Zweiger, Adam and Pari, Jyo and Guo, Han and Aky{\"u}rek, Ekin and Kim, Yoon and Agrawal, Pulkit},
  booktitle = {Advances in Neural Information Processing Systems 38 (NeurIPS 2025)},
  year      = {2025},
  url       = {https://papers.nips.cc/paper_files/paper/2025/hash/6b41e04c41726e2a60e456d0a2b961ab-Abstract-Conference.html}
}

% ============================================================
% §5.3 Benchmark stack
% ============================================================

% OLDKEY: arxiv2310-06770 -> NEWKEY: jimenez2024swebench
@inproceedings{jimenez2024swebench,
  title     = {{SWE}-bench: Can Language Models Resolve Real-World {GitHub} Issues?},
  author    = {Jimenez, Carlos E. and Yang, John and Wettig, Alexander and Yao, Shunyu and Pei, Kexin and Press, Ofir and Narasimhan, Karthik},
  booktitle = {International Conference on Learning Representations},
  year      = {2024}
}

% OLDKEY: arxiv2601-11868 -> NEWKEY: merrill2026terminalbench
% NOTE: task-supplied title ("...Real Terminal Environments") and first author ("Yang, Mingzhe") do
% not match the actual paper at this arXiv ID. Verified true title/authors via arXiv abstract-page
% metadata (85 authors) and confirmed acceptance via proceedings.iclr.cc.
@inproceedings{merrill2026terminalbench,
  title     = {Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces},
  author    = {Merrill, Mike A. and Shaw, Alexander G. and Carlini, Nicholas and Li, Boxuan and Raj, Harsh and Bercovich, Ivan and Shi, Lin and Shin, Jeong Yeon and Walshe, Thomas and Buchanan, E. Kelly and Shen, Junhong and Ye, Guanghao and Lin, Haowei and Poulos, Jason and Wang, Maoyu and Nezhurina, Marianna and Jitsev, Jenia and Lu, Di and Mastromichalakis, Orfeas Menis and Xu, Zhiwei and Chen, Zizhao and Liu, Yue and Zhang, Robert and Chen, Leon Liangyu and Kashyap, Anurag and Uslu, Jan-Lucas and Li, Jeffrey and Wu, Jianbo and Yan, Minghao and Bian, Song and Sharma, Vedang and Sun, Ke and Dillmann, Steven and Anand, Akshay and Lanpouthakoun, Andrew and Koopah, Bardia and Hu, Changran and Guha, Etash and Dreiman, Gabriel H. S. and Zhu, Jiacheng and Krauth, Karl and Zhong, Li and Muennighoff, Niklas and Amanfu, Robert and Tan, Shangyin and Pimpalgaonkar, Shreyas and Aggarwal, Tushar and Lin, Xiangning and Lan, Xin and Zhao, Xuandong and Liang, Yiqing and Wang, Yuanli and Wang, Zilong and Zhou, Changzhi and Heineman, David and Liu, Hange and Trivedi, Harsh and Yang, John and Lin, Junhong and Shetty, Manish and Yang, Michael and Omi, Nabil and Raoof, Negin and Li, Shanda and Zhuo, Terry Yue and Lin, Wuwei and Dai, Yiwei and Wang, Yuxin and Chai, Wenhao and Zhou, Shang and Wahdany, Dariush and She, Ziyu and Hu, Jiaming and Dong, Zhikang and Zhu, Yuxuan and Cui, Sasha and Saiyed, Ahson and Kolbeinsson, Arinbj{\"o}rn and Hu, Jesse and Rytting, Christopher Michael and Marten, Ryan and Wang, Yixin and Dimakis, Alex and Konwinski, Andy and Schmidt, Ludwig},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2026/file/444a3737adaee10d86ad2ef5f74468e6-Paper-Conference.pdf}
}

% OLDKEY: arxiv2407-01502 -> NEWKEY: kapoor2025matter
% NOTE: task-supplied author "Nitarshan Rajkumar" is WRONG — the actual 4th author, confirmed via
% arXiv abstract-page metadata and Princeton's institutional repository, is "Nitya Nadgir".
@article{kapoor2025matter,
  title   = {{AI} Agents That Matter},
  author  = {Kapoor, Sayash and Stroebl, Benedikt and Siegel, Zachary S. and Nadgir, Nitya and Narayanan, Arvind},
  journal = {Transactions on Machine Learning Research},
  year    = {2025}
}

% OLDKEY: arxiv2410-09024 -> NEWKEY: andriushchenko2025agentharm
@inproceedings{andriushchenko2025agentharm,
  title     = {AgentHarm: A Benchmark for Measuring Harmfulness of {LLM} Agents},
  author    = {Andriushchenko, Maksym and Souly, Alexandra and Dziemian, Mateusz and Duenas, Derek and Lin, Maxwell and Wang, Justin and Hendrycks, Dan and Zou, Andy and Kolter, Zico and Fredrikson, Matt and Winsor, Eric and Wynne, Jerome and Gal, Yarin and Davies, Xander},
  booktitle = {The Thirteenth International Conference on Learning Representations},
  year      = {2025},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2025/hash/c493d23af93118975cdbc32cbe7323f5-Abstract-Conference.html}
}

% OLDKEY: arxiv2510-11977 -> NEWKEY: kapoor2026holistic
@inproceedings{kapoor2026holistic,
  title     = {Holistic Agent Leaderboard: The Missing Infrastructure for {AI} Agent Evaluation},
  author    = {Kapoor, Sayash and Stroebl, Benedikt and Kirgis, Peter and Nadgir, Nitya and Siegel, Zachary S. and Wei, Boyi and Xue, Tianci and Chen, Ziru and Chen, Felix and Utpala, Saiteja and Ndzomga, Franck and Oruganty, Dheeraj and Luskin, Sophie and Liu, Kangheng and Yu, Botao and Arora, Amit and Hahm, Dongyoon and Trivedi, Harsh and Sun, Huan and Lee, Juyong and Jin, Tengjun and Mai, Yifan and Zhou, Yifei and Zhu, Yuxuan and Bommasani, Rishi and Kang, Daniel and Song, Dawn and Henderson, Peter and Su, Yu and Liang, Percy and Narayanan, Arvind},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2026/file/a0928f924a344aaebbb7f6cd8d56e34c-Paper-Conference.pdf}
}

% OLDKEY: arxiv2504-01848 -> NEWKEY: starace2025paperbench
@inproceedings{starace2025paperbench,
  title     = {PaperBench: Evaluating {AI}'s Ability to Replicate {AI} Research},
  author    = {Starace, Giulio and Jaffe, Oliver and Sherburn, Dane and Aung, James and Chan, Jun Shern and Maksin, Leon and Dias, Rachel and Mays, Evan and Kinsella, Benjamin and Thompson, Wyatt and Heidecke, Johannes and Glaese, Amelia and Patwardhan, Tejal},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  series    = {Proceedings of Machine Learning Research},
  volume    = {267},
  pages     = {56843--56873},
  year      = {2025},
  url       = {https://proceedings.mlr.press/v267/starace25a.html}
}

% OLDKEY: arxiv2410-07095 -> NEWKEY: chan2025mlebench
@inproceedings{chan2025mlebench,
  title     = {{MLE}-bench: Evaluating Machine Learning Agents on Machine Learning Engineering},
  author    = {Chan, Jun Shern and Chowdhury, Neil and Jaffe, Oliver and Aung, James and Sherburn, Dane and Mays, Evan and Starace, Giulio and Liu, Kevin and Maksin, Leon and Patwardhan, Tejal and Weng, Lilian and Madry, Aleksander},
  booktitle = {International Conference on Learning Representations},
  year      = {2025}
}

% OLDKEY: arxiv2411-15114 -> NEWKEY: wijk2025rebench
% NOTE: title and author list follow the official PMLR v267 record, which lists 22 authors.
% The arXiv version additionally lists Holden Karnofsky and uses the forms "Tao Lin" and
% "Josh Clymer"; the published record takes precedence here.
@inproceedings{wijk2025rebench,
  title     = {{RE}-Bench: Evaluating Frontier {AI} {R\&D} Capabilities of Language Model Agents against Human Experts},
  author    = {Wijk, Hjalmar and Lin, Tao Roa and Becker, Joel and Jawhar, Sami and Parikh, Neev and Broadley, Thomas and Chan, Lawrence and Chen, Michael and Clymer, Joshua M. and Dhyani, Jai and Ericheva, Elena and Garcia, Katharyn and Goodrich, Brian and Jurkovic, Nikola and Kinniment, Megan and Lajko, Aron and Nix, Seraphina and Sato, Lucas Jun Koba and Saunders, William and Taran, Maksym and West, Ben and Barnes, Elizabeth},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  series    = {Proceedings of Machine Learning Research},
  volume    = {267},
  pages     = {66772--66832},
  year      = {2025},
  url       = {https://proceedings.mlr.press/v267/wijk25a.html}
}

% OLDKEY: arxiv2502-10517 -> NEWKEY: ouyang2025kernelbench
% NOTE: task suggested checking NeurIPS 2025 — this is incorrect; the paper was accepted at
% ICML 2025 (PMLR v267), not NeurIPS.
@inproceedings{ouyang2025kernelbench,
  title     = {{KernelBench}: Can {LLMs} Write Efficient {GPU} Kernels?},
  author    = {Ouyang, Anne and Guo, Simon and Arora, Simran and Zhang, Alex L. and Hu, William and R{\'e}, Christopher and Mirhoseini, Azalia},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  series    = {Proceedings of Machine Learning Research},
  volume    = {267},
  pages     = {47356--47415},
  year      = {2025},
  url       = {https://proceedings.mlr.press/v267/ouyang25a.html}
}

% OLDKEY: arxiv2604-08523 -> NEWKEY: zhang2026clawbench
% No venue found (2026-09 re-audit); arXiv preprint.
% NOTE: task-supplied author list was truncated ("and others" implied); full author list
% (28 authors) verified via arXiv abstract-page metadata.
@misc{zhang2026clawbench,
  title         = {ClawBench: Can {AI} Agents Complete Everyday Online Tasks?},
  author        = {Zhang, Yuxuan and Wang, Yubo and Zhu, Yipeng and Du, Penghui and Miao, Junwen and Lu, Xuan and Li, Zhuofeng and Qu, Xingwei and Guo, Zhengkang and Shen, Yuanzhe and Song, Dingjie and Zhou, Han and Zheng, Tuney and Wu, Xian and Yu, Hao and Cai, Songcheng and Lu, Yi and Hao, Yunzhuo and Lei, Minyi and Chen, Liang and Zou, Kai and Yin, Huifeng and Xu, Wendong and Jiang, Dongfu and Nie, Ping and Liu, Jiaheng and Chen, Wenhu and Allen, Kelsey R.},
  year          = {2026},
  eprint        = {2604.08523},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {arXiv:2604.08523}
}

% ============================================================
% Competitor surveys (§1 differentiation)
% ============================================================

% OLDKEY: arxiv2607-13104 -> NEWKEY: ren2026selfimprovements
@misc{ren2026selfimprovements,
  author        = {Zhe Ren and Yimeng Chen and Dandan Guo and Guowei Rong and Tonghui Li and R. B. Xiong and Qingfeng Lan and Wenyi Wang and Li Nanbo and Yibo Yang and Mingchen Zhuge and J{\"u}rgen Schmidhuber},
  title         = {Self-Improvements in Modern Agentic Systems: A Survey},
  year          = {2026},
  eprint        = {2607.13104},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2607.13104}
}

% OLDKEY: arxiv2604-08224 -> NEWKEY: zhou2026externalization
@misc{zhou2026externalization,
  author        = {Chenyu Zhou and Huacan Chai and Wenteng Chen and Zihan Guo and Rong Shan and Yuanyi Song and Tianyi Xu and Yingxuan Yang and Aofan Yu and Weiming Zhang and Congming Zheng and Jiachen Zhu and Zeyu Zheng and Zhuosheng Zhang and Xingyu Lou and Changwang Zhang and Zhihui Fu and Jun Wang and Weiwen Liu and Jianghao Lin and Weinan Zhang},
  title         = {Externalization in {LLM} Agents: A Unified Review of Memory, Skills, Protocols and Harness Engineering},
  year          = {2026},
  eprint        = {2604.08224},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SE},
  url           = {https://arxiv.org/abs/2604.08224}
}

% OLDKEY: arxiv2507-21046 -> NEWKEY: gao2026selfevolving
@article{gao2026selfevolving,
  title   = {A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence},
  author  = {Gao, Huan-ang and Geng, Jiayi and Hua, Wenyue and Hu, Mengkang and Juan, Xinzhe and Liu, Hongzhang and Liu, Shilong and Qiu, Jiahao and Qi, Xuan and Ren, Qihan and Wu, Yiran and Wang, Hongru and Xiao, Han and Zhou, Yuhang and Zhang, Shaokun and Zhang, Jiayi and Xiang, Jinyu and Fang, Yixiong and Zhao, Qiwen and Liu, Dongrui and Qian, Cheng and Wang, Zhenhailong and Hu, Minda and Wang, Huazheng and Wu, Qingyun and Ji, Heng and Wang, Mengdi},
  journal = {Transactions on Machine Learning Research},
  year    = {2026},
  eprint        = {2507.21046},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2507.21046}
}

% OLDKEY: arxiv2508-07407 -> NEWKEY: fang2025comprehensive
@misc{fang2025comprehensive,
  author        = {Jinyuan Fang and Yanwen Peng and Xi Zhang and Yingxu Wang and Xinhao Yi and Guibin Zhang and Yi Xu and Bin Wu and Siwei Liu and Zihao Li and Zhaochun Ren and Nikos Aletras and Xi Wang and Han Zhou and Zaiqiao Meng},
  title         = {A Comprehensive Survey of Self-Evolving {AI} Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems},
  year          = {2025},
  eprint        = {2508.07407},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2508.07407}
}

% OLDKEY: arxiv2404-14387 -> NEWKEY: tao2024survey
% No venue found (2026-09 re-audit), and the arXiv Comments field carries no "accepted at" note:
% no evidence of IJCAI or any other peer-reviewed acceptance. Remains arXiv-only.
@misc{tao2024survey,
  author        = {Zhengwei Tao and Ting-En Lin and Xiancai Chen and Hangyu Li and Yuchuan Wu and Yongbin Li and Zhi Jin and Fei Huang and Dacheng Tao and Jingren Zhou},
  title         = {A Survey on Self-Evolution of Large Language Models},
  year          = {2024},
  eprint        = {2404.14387},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2404.14387}
}

% OLDKEY: arxiv2507-13334 -> NEWKEY: mei2025survey
@misc{mei2025survey,
  author        = {Lingrui Mei and Jiayu Yao and Yuyao Ge and Yiwei Wang and Baolong Bi and Yujun Cai and Jiazhi Liu and Mingyu Li and Zhong-Zhi Li and Duzhen Zhang and Chenlin Zhou and Jiayi Mao and Tianze Xia and Jiafeng Guo and Shenghua Liu},
  title         = {A Survey of Context Engineering for Large Language Models},
  year          = {2025},
  eprint        = {2507.13334},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2507.13334}
}

% ============================================================
% §7 Safety and risk anchors
% ============================================================

% OLDKEY: arxiv2209-13085 -> NEWKEY: skalse2022defining
% NOTE: the official NeurIPS proceedings site (papers.nips.cc) titles the camera-ready
% "Defining and Characterizing Reward Gaming" and gives pages 9460-9471, and lists the second
% author as "Nikolaus Howe"; the arXiv preprint and the Crossref record for the Curran volume
% both still carry the original title "...Reward Hacking". The proceedings-site form is used here.
@inproceedings{skalse2022defining,
  author    = {Joar Skalse and Nikolaus Howe and Dmitrii Krasheninnikov and David Krueger},
  title     = {Defining and Characterizing Reward Gaming},
  booktitle = {Advances in Neural Information Processing Systems 35 (NeurIPS 2022)},
  volume    = {35},
  pages     = {9460--9471},
  publisher = {Curran Associates, Inc.},
  year      = {2022},
  url       = {https://papers.nips.cc/paper_files/paper/2022/hash/3d719fee332caa23d5038b8a90e81796-Abstract-Conference.html},
  note      = {Originally released as arXiv:2209.13085 under the title ``Defining and Characterizing Reward Hacking''}
}

% OLDKEY: arxiv2210-10760 -> NEWKEY: gao2023scaling
@inproceedings{gao2023scaling,
  author    = {Leo Gao and John Schulman and Jacob Hilton},
  title     = {Scaling Laws for Reward Model Overoptimization},
  booktitle = {Proceedings of the 40th International Conference on Machine Learning (ICML 2023)},
  volume    = {202},
  series    = {Proceedings of Machine Learning Research},
  pages     = {10835--10866},
  publisher = {PMLR},
  year      = {2023},
  note      = {Originally released as arXiv:2210.10760}
}

% OLDKEY: arxiv2406-10162 -> NEWKEY: denison2024sycophancy
% Checked NeurIPS 2024 proceedings listing directly: paper does not appear there. Remains arXiv-only.
% Corrected title (arXiv metadata): "...Investigating Reward-Tampering in Large Language Models" (task text had "Reward Tampering in Language Models").
@misc{denison2024sycophancy,
  author        = {Carson Denison and Monte MacDiarmid and Fazl Barez and David Duvenaud and Shauna Kravec and Samuel Marks and Nicholas Schiefer and Ryan Soklaski and Alex Tamkin and Jared Kaplan and Buck Shlegeris and Samuel R. Bowman and Ethan Perez and Evan Hubinger},
  title         = {Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models},
  year          = {2024},
  eprint        = {2406.10162},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2406.10162}
}

% OLDKEY: arxiv2503-11926 -> NEWKEY: baker2025monitoring
@misc{baker2025monitoring,
  author        = {Bowen Baker and Joost Huizinga and Leo Gao and Zehao Dou and Melody Y. Guan and Aleksander Madry and Wojciech Zaremba and Jakub Pachocki and David Farhi},
  title         = {Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation},
  year          = {2025},
  eprint        = {2503.11926},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2503.11926}
}

% OLDKEY: arxiv2310-01798 -> NEWKEY: huang2024large
% NOTE: the previous version carried "pages = {32808--32824}", which has no source -- ICLR
% proceedings papers are not paginated. Removed and replaced with the official proceedings URL.
@inproceedings{huang2024large,
  author    = {Jie Huang and Xinyun Chen and Swaroop Mishra and Huaixiu Steven Zheng and Adams Wei Yu and Xinying Song and Denny Zhou},
  title     = {Large Language Models Cannot Self-Correct Reasoning Yet},
  booktitle = {The Twelfth International Conference on Learning Representations},
  year      = {2024},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2024/hash/8b4add8b0aa8749d80a34ca5d941c355-Abstract-Conference.html},
  note      = {Originally released as arXiv:2310.01798}
}

% ============================================================
% §2 PAC/ZO theory additional anchors
% ============================================================



% ============================================================
% Other commonly-cited works
% ============================================================






% OLDKEY: arxiv2406-18532 -> NEWKEY: ou2025symbolic
% Published version located in the 2026-09-10 re-audit (Crossref). NOTE: in the journal version the
% first author is Yixin Ou, not Wangchunshu Zhou -- the author order differs from the arXiv preprint,
% which is why the citation key changed from zhou2024symbolic to ou2025symbolic.
@article{ou2025symbolic,
  author  = {Yixin Ou and Wangchunshu Zhou and Shengwei Ding and Long Li and Jialong Wu and Tiannan Wang and Jiamin Chen and Shuai Wang and Xiaohua Xu and Ningyu Zhang and Huajun Chen and Yuchen Eleanor Jiang},
  title   = {Symbolic learning enables self-evolving agents},
  journal = {AI Open},
  volume  = {6},
  pages   = {314--322},
  year    = {2025},
  doi     = {10.1016/j.aiopen.2025.11.004},
  note    = {Originally released as arXiv:2406.18532}
}

% OLDKEY: arxiv2502-02533 -> NEWKEY: zhou2026multiagent
% Accepted at ICLR 2026; confirmed in the official proceedings index (2026-09-10 re-audit).
% The previous version of this entry incorrectly recorded the paper as arXiv-only.
@inproceedings{zhou2026multiagent,
  author    = {Han Zhou and Xingchen Wan and Ruoxi Sun and Hamid Palangi and Shariq Iqbal and Ivan Vuli{\'c} and Anna Korhonen and Sercan {\"O}. Ar{\i}k},
  title     = {Multi-Agent Design: Optimizing Agents with Better Prompts and Topologies},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2026/hash/1ab4e0e8f35078109bc78d7b465d306f-Abstract-Conference.html},
  note      = {Originally released as arXiv:2502.02533}
}


% OLDKEY: arxiv2608-05144 -> NEWKEY: li2026argus
% NOTE: task listed author as "Argus Team" (organizational), but the arXiv record itself lists
% 27 individual named authors led by Boxiu Li -- used the actual arXiv author list, not fabricated org name.
@misc{li2026argus,
  author        = {Boxiu Li and Zimo Wen and Yijia Fan and Chuan Wen and Fan Yang and Hangxi Guo and Jiaao Wu and Jiachen Zhang and Junxiang Lei and Mukai Li and Ruize Tang and Runjing Gu and Shibo Hu and Sihan Chen and Sufeng Guo and Wanbo Zhang and Xian Zhang and Xiaoyu Chen and Xuanhe Zhou and Xuyao Huang and Yifei Gao and Yifei Shen and Yilin Chen and Yuheng Wu and Yuzhe Zhang and Zelong Zhao and Zhijie Deng},
  title         = {Argus: A General-Purpose Agentic Reasoning Runtime for Long-Horizon Tasks},
  year          = {2026},
  eprint        = {2608.05144},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2608.05144}
}

% OLDKEY: arxiv2502-13138 -> NEWKEY: jiang2025aide
% No venue found (2026-09 re-audit); kept as an arXiv preprint.
@misc{jiang2025aide,
  author        = {Zhengyao Jiang and Dominik Schmidt and Dhruv Srikanth and Dixing Xu and Ian Kaplan and Deniss Jacenko and Yuxiang Wu},
  title         = {{AIDE}: AI-Driven Exploration in the Space of Code},
  year          = {2025},
  eprint        = {2502.13138},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2502.13138}
}

% OLDKEY: arxiv2504-08066 -> NEWKEY: yamada2025scientist
% Title itself ("Workshop-Level") signals non-peer-reviewed scope; no venue found.
@misc{yamada2025scientist,
  author        = {Yutaro Yamada and Robert Tjarko Lange and Cong Lu and Shengran Hu and Chris Lu and Jakob Foerster and Jeff Clune and David Ha},
  title         = {The {AI} Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search},
  year          = {2025},
  eprint        = {2504.08066},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2504.08066}
}

% OLDKEY: arxiv2601-03315 -> NEWKEY: trehan2026why
@misc{trehan2026why,
  author        = {Dhruv Trehan and Paras Chopra},
  title         = {Why {LLMs} Aren't Scientists Yet: Lessons from Four Autonomous Research Attempts},
  year          = {2026},
  eprint        = {2601.03315},
  archivePrefix = {arXiv},
  primaryClass  = {cs.LG},
  url           = {https://arxiv.org/abs/2601.03315}
}

% OLDKEY: arxiv2511-16072 -> NEWKEY: bubeck2025early
% Corrected title case per arXiv metadata: "Early science acceleration experiments with GPT-5"
% (task text used title case). Microsoft/OpenAI technical report; no peer-reviewed venue found.
@misc{bubeck2025early,
  author        = {S{\'e}bastien Bubeck and Christian Coester and Ronen Eldan and Timothy Gowers and Yin Tat Lee and Alexandru Lupsasca and Mehtaab Sawhney and Robert Scherrer and Mark Sellke and Brian K. Spears and Derya Unutmaz and Kevin Weil and Steven Yin and Nikita Zhivotovskiy},
  title         = {Early science acceleration experiments with {GPT-5}},
  year          = {2025},
  eprint        = {2511.16072},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2511.16072}
}

% OLDKEY: arxiv2605-27276 -> NEWKEY: hebbar2026sia
@misc{hebbar2026sia,
  author        = {Prannay Hebbar and Yogendra Manawat and Samuel Verboomen and Alesia Ivanova and Selvam Palanimalai and Kunal Bhatia and Vignesh Baskaran},
  title         = {{SIA}: Self Improving {AI} with Harness \& Weight Updates},
  year          = {2026},
  eprint        = {2605.27276},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2605.27276}
}

% OLDKEY: arxiv2606-20333 -> NEWKEY: tao2026softskill
@misc{tao2026softskill,
  title={{SoftSkill}: Behavioral Compression for Contextual Adaptation},
  author={Xijia Tao and Yihua Teng and Xinyu Fu and Ziru Liu and Kecheng Chen and Yuzhi Zhao and Suiyun Zhang and Rui Liu and Lingpeng Kong},
  year={2026},
  eprint={2606.20333},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url           = {https://arxiv.org/abs/2606.20333}
}

% OLDKEY: arxiv2603-19461 -> NEWKEY: zhang2026hyperagents
% No venue found (2026-09 re-audit); in particular the paper is NOT in the ICLR 2026 proceedings
% index, despite secondary sources claiming so. Kept as an arXiv preprint.
@misc{zhang2026hyperagents,
  title={HyperAgents},
  author={Jenny Zhang and Bingchen Zhao and Wannan Yang and Jakob Foerster and Jeff Clune and Minqi Jiang and Sam Devlin and Tatiana Shavrina},
  year={2026},
  eprint={2603.19461},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url           = {https://arxiv.org/abs/2603.19461}
}

% OLDKEY: arxiv2604-01658 -> NEWKEY: qu2026coral
@inproceedings{qu2026coral,
  title={{CORAL}: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery},
  author={Ao Qu and Han Zheng and Zijian Zhou and Yihao Yan and Yihong Tang and Shao Yong Ong and Fenglu Hong and Kaichen Zhou and Chonghe Jiang and Minwei Kong and Jiacheng Zhu and Xuan Jiang and Sirui Li and Cathy Wu and Bryan Kian Hsiang Low and Jinhua Zhao and Paul Pu Liang},
  booktitle={The Third Conference on Language Modeling (COLM 2026)},
  year={2026},
  note={arXiv:2604.01658}
}

% OLDKEY: arxiv2504-07079 -> NEWKEY: zheng2025skillweaver
% No venue found (2026-09 re-audit); kept as an arXiv preprint.
@misc{zheng2025skillweaver,
  title={SkillWeaver: Web Agents can Self-Improve by Discovering and Honing Skills},
  author={Boyuan Zheng and Michael Y. Fatemi and Xiaolong Jin and Zora Zhiruo Wang and Apurva Gandhi and Yueqi Song and Yu Gu and Jayanth Srinivasa and Gaowen Liu and Graham Neubig and Yu Su},
  year={2025},
  eprint={2504.07079},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url           = {https://arxiv.org/abs/2504.07079}
}


% OLDKEY: lenat1983eurisko -> NEWKEY: lenat1983eurisko
@article{lenat1983eurisko,
  title   = {{EURISKO}: A Program That Learns New Heuristics and Domain Concepts},
  author  = {Lenat, Douglas B.},
  journal = {Artificial Intelligence},
  volume  = {21},
  number  = {1--2},
  pages   = {61--98},
  year    = {1983}
}

% OLDKEY: arxiv2406-12045 -> NEWKEY: yao2025taubench
@inproceedings{yao2025taubench,
  title={$\tau$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains},
  author={Shunyu Yao and Noah Shinn and Pedram Razavi and Karthik R. Narasimhan},
  booktitle={The Thirteenth International Conference on Learning Representations (ICLR 2025)},
  year={2025},
  note={arXiv:2406.12045}
}

% Added 2026-09-26: source of the pass-at-k unbiased estimator that tau-bench restates.
@misc{chen2021codex,
  title         = {Evaluating Large Language Models Trained on Code},
  author        = {Mark Chen and Jerry Tworek and Heewoo Jun and Qiming Yuan and Henrique Ponde de Oliveira Pinto and Jared Kaplan and Harri Edwards and Yuri Burda and Nicholas Joseph and Greg Brockman and others},
  year          = {2021},
  eprint        = {2107.03374},
  archivePrefix = {arXiv},
  primaryClass  = {cs.LG},
  note          = {arXiv preprint},
  url           = {https://arxiv.org/abs/2107.03374}
}

% Added 2026-09-26: the evaluation suite used in MOSS's case study (MOSS writes it "claweval").
@misc{ye2026claweval,
  title         = {Claw-Eval: Towards Trustworthy Evaluation of Autonomous Agents},
  author        = {Bowen Ye and Rang Li and Qibin Yang and Yuanxin Liu and Linli Yao and Hanglong Lv and Zhihui Xie and Chenxin An and Lei Li and Lingpeng Kong and Qi Liu and Zhifang Sui and Tong Yang},
  year          = {2026},
  eprint        = {2604.06132},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  note          = {arXiv preprint},
  url           = {https://arxiv.org/abs/2604.06132}
}

% OLDKEY: arxiv2406-13352 -> NEWKEY: debenedetti2024agentdojo
@inproceedings{debenedetti2024agentdojo,
  title={{AgentDojo}: A Dynamic Environment to Evaluate Prompt Injection Attacks and Defenses for {LLM} Agents},
  author={Edoardo Debenedetti and Jie Zhang and Mislav Balunovi{\'c} and Luca Beurer-Kellner and Marc Fischer and Florian Tram{\`e}r},
  booktitle={Advances in Neural Information Processing Systems 37 (NeurIPS 2024), Datasets and Benchmarks Track},
  year={2024},
  note={OpenReview: https://openreview.net/forum?id=m1YYAQjO3w}
}

% OLDKEY: arxiv2404-07972 -> NEWKEY: xie2024osworld
@inproceedings{xie2024osworld,
  title={{OSWorld}: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments},
  author={Tianbao Xie and Danyang Zhang and Jixuan Chen and Xiaochuan Li and Siheng Zhao and Ruisheng Cao and Toh Jing Hua and Zhoujun Cheng and Dongchan Shin and Fangyu Lei and Yitao Liu and Yiheng Xu and Shuyan Zhou and Silvio Savarese and Caiming Xiong and Victor Zhong and Tao Yu},
  booktitle={Advances in Neural Information Processing Systems 37 (NeurIPS 2024), Datasets and Benchmarks Track},
  year={2024}
}

% OLDKEY: arxiv2412-14470 -> NEWKEY: zhang2024agentsafetybench
@misc{zhang2024agentsafetybench,
  title={{Agent-SafetyBench}: Evaluating the Safety of {LLM} Agents},
  author={Zhexin Zhang and Shiyao Cui and Yida Lu and Jingzhuo Zhou and Junxiao Yang and Hongning Wang and Minlie Huang},
  year={2024},
  eprint={2412.14470},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  url           = {https://arxiv.org/abs/2412.14470}
}

% OLDKEY: metr2024 -> NEWKEY: metr2024
@misc{metr2024,
  title        = {Autonomy Evaluation Resources},
  author       = {{METR}},
  year         = {2024},
  howpublished = {\url{https://metr.org/blog/2024-03-13-autonomy-evaluation-resources/}},
  note         = {Long-horizon agent evaluation framework, announced 2024-03-13}
}

% OLDKEY: godel1931formal -> NEWKEY: godel1931formal
@article{godel1931formal,
  title   = {{\"U}ber formal unentscheidbare S{\"a}tze der Principia Mathematica und verwandter Systeme I},
  author  = {G{\"o}del, Kurt},
  journal = {Monatshefte f{\"u}r Mathematik und Physik},
  volume  = {38},
  pages   = {173--198},
  year    = {1931}
}

% --- Added 2026-09-09: v2 literature audit (LITERATURE-REVIEW_v2.md) ---

% OLDKEY: arxiv2407-04549 -> NEWKEY: pan2024spontaneous
@misc{pan2024spontaneous,
  title={Spontaneous Reward Hacking in Iterative Self-Refinement},
  author={Jane Pan and He He and Samuel R. Bowman and Shi Feng},
  year={2024},
  eprint={2407.04549},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  url           = {https://arxiv.org/abs/2407.04549}
}

% OLDKEY: arxiv2410-08164 -> NEWKEY: agashe2025agent
@inproceedings{agashe2025agent,
  title={Agent S: An Open Agentic Framework that Uses Computers Like a Human},
  author={Saaket Agashe and Jiuzhou Han and Shuyu Gan and Jiachen Yang and Ang Li and Xin Eric Wang},
  booktitle={The Thirteenth International Conference on Learning Representations (ICLR 2025)},
  year={2025}
}

% OLDKEY: arxiv2501-11733 -> NEWKEY: wang2025mobileagente
@misc{wang2025mobileagente,
  title={Mobile-Agent-E: Self-Evolving Mobile Assistant for Complex Tasks},
  author={Zhenhailong Wang and Haiyang Xu and Junyang Wang and Xi Zhang and Ming Yan and Ji Zhang and Fei Huang and Heng Ji},
  year={2025},
  eprint={2501.11733},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  url           = {https://arxiv.org/abs/2501.11733}
}

% OLDKEY: arxiv2503-02268 -> NEWKEY: jiang2025appagentx
@misc{jiang2025appagentx,
  title={{AppAgentX}: Evolving {GUI} Agents as Proficient Smartphone Users},
  author={Wenjia Jiang and Yangyang Zhuang and Chenxi Song and Xu Yang and Joey Tianyi Zhou and Chi Zhang},
  year={2025},
  eprint={2503.02268},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url           = {https://arxiv.org/abs/2503.02268}
}

% OLDKEY: arxiv2504-06821 -> NEWKEY: wang2025inducing
% Accepted at COLM 2025; confirmed on colm.cc (2026-09-10 re-audit). COLM does not paginate
% its proceedings, so no page range is given.
@inproceedings{wang2025inducing,
  title     = {Inducing Programmatic Skills for Agentic Tasks},
  author    = {Zora Zhiruo Wang and Apurva Gandhi and Graham Neubig and Daniel Fried},
  booktitle = {Second Conference on Language Modeling (COLM 2025)},
  year      = {2025},
  url       = {https://colm.cc/virtual/2025/poster/533},
  note      = {Originally released as arXiv:2504.06821}
}

% OLDKEY: arxiv2505-16827 -> NEWKEY: xie2025guiexplorer
% Venue, volume and pages confirmed via Crossref/ACL Anthology (2026-09-10 re-audit).
@inproceedings{xie2025guiexplorer,
  title     = {{GUI}-explorer: Autonomous Exploration and Mining of Transition-aware Knowledge for {GUI} Agent},
  author    = {Bin Xie and Rui Shao and Gongwei Chen and Kaiwen Zhou and Yinchuan Li and Jie Liu and Min Zhang and Liqiang Nie},
  booktitle = {Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  month     = jul,
  year      = {2025},
  address   = {Vienna, Austria},
  publisher = {Association for Computational Linguistics},
  pages     = {5650--5667},
  doi       = {10.18653/v1/2025.acl-long.282},
  url       = {https://aclanthology.org/2025.acl-long.282/}
}

% OLDKEY: arxiv2505-20286 -> NEWKEY: qiu2025alita
% No venue found (2026-09 re-audit); kept as an arXiv preprint.
@misc{qiu2025alita,
  title={Alita: Generalist Agent Enabling Scalable Agentic Reasoning with Minimal Predefinition and Maximal Self-Evolution},
  author={Jiahao Qiu and Xuan Qi and Tongcheng Zhang and Xinzhe Juan and Jiacheng Guo and Yifu Lu and Yimin Wang and Zixin Yao and Qihan Ren and Xun Jiang and Xing Zhou and Dongrui Liu and Ling Yang and Yue Wu and Kaixuan Huang and Shilong Liu and Hongru Wang and Mengdi Wang},
  year={2025},
  eprint={2505.20286},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url           = {https://arxiv.org/abs/2505.20286}
}

% OLDKEY: arxiv2508-04700 -> NEWKEY: sun2026seagent
% Accepted at ICML 2026; confirmed on icml.cc/virtual/2026 (2026-09-10 re-audit).
% PMLR volume/pages not yet assigned, so they are omitted rather than invented.
@inproceedings{sun2026seagent,
  title={{SEAgent}: Self-Evolving Computer Use Agent with Autonomous Learning from Experience},
  author={Zeyi Sun and Ziyu Liu and Yuhang Zang and Yuhang Cao and Xiaoyi Dong and Tong Wu and Dahua Lin and Jiaqi Wang},
  booktitle={Proceedings of the 43rd International Conference on Machine Learning (ICML 2026)},
  year={2026},
  url={https://icml.cc/virtual/2026/poster/65711},
  note={arXiv:2508.04700}
}

% OLDKEY: arxiv2509-18597 -> NEWKEY: meng2025growing
% No venue found (2026-09 re-audit); kept as an arXiv preprint.
@misc{meng2025growing,
  title={Robotic Ultra-Long-Horizon Manipulation Skills via Human-guided Lifelong Code Generation},
  author={Yuan Meng and Zhenguo Sun and Max Fest and Xiangtong Yao and Xukun Li and Zhenshan Bing and Alois Knoll},
  year={2025},
  eprint={2509.18597},
  archivePrefix={arXiv},
  primaryClass={cs.RO},
  url           = {https://arxiv.org/abs/2509.18597}
}

% OLDKEY: arxiv2509-24160 -> NEWKEY: kagaya2025memory
% No venue found (2026-09 re-audit) — kept as an arXiv preprint.
@misc{kagaya2025memory,
  title={Memory Transfer Planning: {LLM}-driven Context-Aware Code Adaptation for Robot Manipulation},
  author={Tomoyuki Kagaya and Subramanian Lakshmi and Yuxuan Lou and Thong Jing Yuan and Jayashree Karlekar and Sugiri Pranata and Natsuki Murakami and Akira Kinose and Yang You},
  year={2025},
  eprint={2509.24160},
  archivePrefix={arXiv},
  primaryClass={cs.RO},
  url           = {https://arxiv.org/abs/2509.24160}
}

% OLDKEY: arxiv2510-04374 -> NEWKEY: patwardhan2026gdpval
% Accepted at ICLR 2026; confirmed in the official proceedings index (2026-09-10 re-audit).
% Author names and order follow the ICLR record, which differs slightly from the arXiv byline
% ("Simon Fishman" / "Natalie Kim", and Patrick Chao appears later in the list).
@inproceedings{patwardhan2026gdpval,
  title     = {{GDPval}: Evaluating {AI} Model Performance on Real-World Economically Valuable Tasks},
  author    = {Tejal Patwardhan and Rachel Dias and Elizabeth Proehl and Grace Kim and Michele Wang and Olivia Watkins and Simon Fishman and Marwan Aljubeh and Phoebe Thacker and Laurance Fauconnet and Natalie Kim and Samuel Miserendino and Gildas Chabot and David Li and Patrick Chao and Michael Sharman and Alexandra Barr and Amelia Glaese and Jerry Tworek},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://proceedings.iclr.cc/paper_files/paper/2026/hash/290c2430f91912204f30bbcc990fff1d-Abstract-Conference.html},
  note      = {Originally released as arXiv:2510.04374}
}

% OLDKEY: arxiv2511-12997 -> NEWKEY: liu2025webcoach
@misc{liu2025webcoach,
      title={WebCoach: Self-Evolving Web Agents with Cross-Session Memory Guidance},
      author={Genglin Liu and Shijie Geng and Sha Li and Hejie Cui and Sarah Zhang and Xin Liu and Tianyi Liu},
      year={2025},
      eprint={2511.12997},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2511.12997}
}

% OLDKEY: arxiv2603-12056 -> NEWKEY: jiang2026xskill
% Accepted at ICML 2026; confirmed on icml.cc/virtual/2026 (2026-09-10 re-audit).
@inproceedings{jiang2026xskill,
      title={XSkill: Continual Learning from Experience and Skills in Multimodal Agents},
      author={Guanyu Jiang and Zhaochen Su and Xiaoye Qu and Yi R. Fung},
      booktitle={Proceedings of the 43rd International Conference on Machine Learning (ICML 2026)},
      year={2026},
      url={https://icml.cc/virtual/2026/poster/65729},
      note={PMLR volume/pages not yet assigned; arXiv:2603.12056}
}

% OLDKEY: arxiv2603-13131 -> NEWKEY: xie2026mineevolve
@misc{xie2026mineevolve,
      title={{MineEvolve}: Self-Evolution with Accumulated Knowledge for Long-Horizon Embodied {Minecraft} Agents},
      author={Zhengwei Xie and Zhisheng Chen and Ziyan Weng and Jinhan Li and Chenglong Li and Zikai Xiao and Jingwei Song and Jinhao Jing and Vireo Zhang and Kun Wang},
      year={2026},
      eprint={2603.13131},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2603.13131}
}

% OLDKEY: arxiv2605-22794 -> NEWKEY: cai2026moss
@misc{cai2026moss,
      title={{MOSS}: Self-Evolution through Source-Level Rewriting in Autonomous Agent Systems},
      author={Qianshu Cai and Yonggang Zhang and Xianzhang Jia and Huajiang Zheng and Wei Xue and Jun Song and Xinmei Tian and Yike Guo},
      year={2026},
      eprint={2605.22794},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2605.22794}
}

% OLDKEY: arxiv2605-26340 -> NEWKEY: meng2026scientistone
@misc{meng2026scientistone,
      title={{ScientistOne}: Towards Human-Level Autonomous Research via Chain-of-Evidence},
      author={Rui Meng and Bhavana Dalvi Mishra and Jiefeng Chen and Chun-Liang Li and Palash Goyal and Mihir Parmar and Yiwen Song and Yale Song and Rajarishi Sinha and Parthasarathy Ranganathan and Burak Gokturk and Jinsung Yoon and Tomas Pfister},
      year={2026},
      eprint={2605.26340},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2605.26340}
}

% OLDKEY: arxiv2606-04329 -> NEWKEY: dash2026untrusted
@misc{dash2026untrusted,
      title={From Untrusted Input to Trusted Memory: A Systematic Study of Memory Poisoning Attacks in {LLM} Agents},
      author={Pritam Dash and Tongyu Ge and Aditi Jain and Tanmay Shah and Zhiwei Shang},
      year={2026},
      eprint={2606.04329},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2606.04329}
}

% OLDKEY: arxiv2606-31121 -> NEWKEY: chen2026prologue
@misc{chen2026prologue,
      title={The Past Is Prologue: A Plug-in Controller for Selective Updates in Sequentially Evolving {LLM} Memory},
      author={Zihan Chen and Songwei Dong and Chengshuai Shi and Peng Wang and Song Wang and Cong Shen and Jundong Li},
      year={2026},
      eprint={2606.31121},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2606.31121}
}

% OLDKEY: arxiv2607-14611 -> NEWKEY: gadgil2026bad
@misc{gadgil2026bad,
      title={Bad Memory: Evaluating Prompt Injection Risks from Memory in Agentic Systems},
      author={Soham Gadgil and David Alexander and Sai Sunku and Franziska Roesner},
      year={2026},
      eprint={2607.14611},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2607.14611}
}

% OLDKEY: arxiv2608-01149 -> NEWKEY: yang2026pathbench
@misc{yang2026pathbench,
      title={{PATH-Bench}: Path-Dependent Evaluation of Lifelong Agents},
      author={Xidong Yang and Xingyi Zhang and Wenhao Li and Wenyan Liu and Junjie Sheng and Yun Hua and Wei Yin and Tao Fang and Chuyun Shen and Xiangfeng Wang},
      year={2026},
      eprint={2608.01149},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2608.01149}
}

% OLDKEY: arxiv2608-03392 -> NEWKEY: zhou2026selfevolving
@misc{zhou2026selfevolving,
      title={Self-Evolving Coding Agents},
      author={Hao Zhou and Haichuan Hu and Tianyu Luo and Ye Shang and Chunrong Fang and Zhenyu Chen and Liang Xiao and Quanjun Zhang},
      year={2026},
      eprint={2608.03392},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      note={arXiv:2608.03392}
}

% OLDKEY: arxiv2608-08311 -> NEWKEY: razzhigaev2026ouroboros
@misc{razzhigaev2026ouroboros,
      title={Ouroboros: A Self-Developing Frontier Coding Agent with Reviewed Core Evolution},
      author={Anton Razzhigaev and Andrei Gritsaev and Andrei Kaznacheev and Nikita Dragunov and Roman Yampolskiy and Andrei Kuznetsov},
      year={2026},
      eprint={2608.08311},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      note={arXiv:2608.08311}
}

% OLDKEY: arxiv2608-25512 -> NEWKEY: shi2026programming
@misc{shi2026programming,
      title={A Programming Paradigm for Spatiotemporal Composability},
      author={Yifan Shi and Wei Zhang and Tianyi Cui},
      year={2026},
      eprint={2608.25512},
      archivePrefix={arXiv},
      primaryClass={cs.PL},
      note={arXiv:2608.25512}
}

% OLDKEY: arxiv2608-30429 -> NEWKEY: kim2026evoskill
% The arXiv "Comments" field states "Accepted to EMNLP 2026". EMNLP 2026 proceedings are not
% published yet, so this stays a preprint entry rather than gaining invented page numbers.
@misc{kim2026evoskill,
      title={{EvoSkill} Injection: Red-Teaming Autonomous Skill Generation and Evolution in Self-Evolving Agents},
      author={Doyun Kim and Chanwoo Kim and Sugyeong Eo and Yeo-Chan Yoon and Chanjun Park},
      year={2026},
      eprint={2608.30429},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      note={Accepted to EMNLP 2026 (per the arXiv comments field); arXiv:2608.30429}
}

% OLDKEY: anthropic2025agentskills -> NEWKEY: anthropic2025agentskills
@misc{anthropic2025agentskills,
  title        = {Equipping Agents for the Real World with Agent Skills},
  author       = {{Anthropic}},
  year         = {2025},
  howpublished = {\url{https://www.anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills}}
}

% OLDKEY: anthropic2026claudecodememory -> NEWKEY: anthropic2026claudecodememory
@misc{anthropic2026claudecodememory,
  title        = {How {Claude} Remembers Your Project},
  author       = {{Anthropic}},
  year         = {2026},
  howpublished = {\url{https://code.claude.com/docs/en/memory}},
  note         = {Claude Code documentation}
}

% OLDKEY: anthropic2025contexteng -> NEWKEY: anthropic2025contexteng
@misc{anthropic2025contexteng,
  title        = {Effective Context Engineering for {AI} Agents},
  author       = {{Anthropic}},
  year         = {2025},
  howpublished = {\url{https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents}}
}

% OLDKEY: anthropic2025writingtools -> NEWKEY: anthropic2025writingtools
@misc{anthropic2025writingtools,
  title        = {Writing Effective Tools for {AI} Agents{\textemdash}Using {AI} Agents},
  author       = {{Anthropic}},
  year         = {2025},
  howpublished = {\url{https://www.anthropic.com/engineering/writing-tools-for-agents}}
}

% OLDKEY: anthropic2024buildingagents -> NEWKEY: anthropic2024buildingagents
@misc{anthropic2024buildingagents,
  title        = {Building Effective {AI} Agents},
  author       = {Schluntz, Erik and Zhang, Barry},
  year         = {2024},
  howpublished = {\url{https://www.anthropic.com/engineering/building-effective-agents}},
  note         = {Anthropic engineering blog, December 2024}
}

% OLDKEY: sakana2025dgmblog -> NEWKEY: sakana2025dgmblog
@misc{sakana2025dgmblog,
  title        = {The Darwin G{\"o}del Machine: {AI} that Improves Itself by Rewriting Its Own Code},
  author       = {{Sakana AI}},
  year         = {2025},
  howpublished = {\url{https://sakana.ai/dgm/}},
  note         = {Blog post, May 2025}
}

% OLDKEY: deepmind2025alphaevolveblog -> NEWKEY: deepmind2025alphaevolveblog
@misc{deepmind2025alphaevolveblog,
  title        = {AlphaEvolve: A Gemini-Powered Coding Agent for Designing Advanced Algorithms},
  author       = {{Google DeepMind}},
  year         = {2025},
  howpublished = {\url{https://deepmind.google/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/}},
  note         = {Blog post, 14 May 2025}
}

% OLDKEY: sakana2024aiscientistblog -> NEWKEY: sakana2024aiscientistblog
@misc{sakana2024aiscientistblog,
  title        = {The {AI} Scientist: Towards Fully Automated Open-Ended Scientific Discovery},
  author       = {{Sakana AI}},
  year         = {2024},
  howpublished = {\url{https://sakana.ai/ai-scientist/}},
  note         = {Blog post, 13 August 2024}
}

% OLDKEY: willison2025claudeskills -> NEWKEY: willison2025claudeskills
@misc{willison2025claudeskills,
  title        = {Claude Skills Are Awesome, Maybe a Bigger Deal than {MCP}},
  author       = {Willison, Simon},
  year         = {2025},
  howpublished = {\url{https://simonwillison.net/2025/Oct/16/claude-skills/}},
  note         = {Blog post, 16 October 2025}
}

% OLDKEY: langchain2024memory -> NEWKEY: langchain2024memory
@misc{langchain2024memory,
  title        = {Memory for Agents},
  author       = {Chase, Harrison},
  year         = {2024},
  howpublished = {\url{https://www.langchain.com/blog/memory-for-agents}},
  note         = {LangChain blog post, October 2024}
}

% OLDKEY: deepseek2026harness -> NEWKEY: deepseek2026harness
@misc{deepseek2026harness,
  title        = {{DeepSeek} Harness Developer Preview},
  author       = {{DeepSeek-AI}},
  year         = {2026},
  howpublished = {\url{https://www.deepseek.com/harness/en/}},
  note         = {Product/developer-preview page}
}

@misc{psiVision,
  title        = {The {AI}-Native Laboratory Fueling the Physics Takeoff},
  author       = {{PSI}},
  year         = {n.d.},
  howpublished = {\url{https://psi.inc/vision/}},
  note         = {Company vision and architecture page; accessed 22 September 2026}
}

% --- Added 2026-09-09: six pending adjacent papers verified (LITERATURE-REVIEW_v2.md tail) ---
% OLDKEY: arxiv2404-06474 -> NEWKEY: pan2024autonomous
@inproceedings{pan2024autonomous,
      title={Autonomous Evaluation and Refinement of Digital Agents},
      author={Jiayi Pan and Yichi Zhang and Nicholas Tomlin and Yifei Zhou and Sergey Levine and Alane Suhr},
      booktitle={First Conference on Language Modeling (COLM 2024)},
      year={2024},
      url={https://openreview.net/forum?id=NPAQ6FKSmK}
}

% OLDKEY: arxiv2504-00906 -> NEWKEY: agashe2025compositional
@inproceedings{agashe2025compositional,
      title={Agent {S2}: A Compositional Generalist-Specialist Framework for Computer Use Agents},
      author={Saaket Agashe and Kyle Wong and Vincent Tu and Jiachen Yang and Ang Li and Xin Eric Wang},
      booktitle={Second Conference on Language Modeling (COLM 2025)},
      year={2025}
}

% OLDKEY: arxiv2504-08942 -> NEWKEY: lu2025agentrewardbench
% Accepted at COLM 2025; confirmed on colm.cc (2026-09-10 re-audit). The previous version of this
% entry recorded the paper as arXiv-only after checking the EMNLP 2025 list -- the wrong venue.
@inproceedings{lu2025agentrewardbench,
  title     = {{AgentRewardBench}: Evaluating Automatic Evaluations of Web Agent Trajectories},
  author    = {Xing Han L\`u and Amirhossein Kazemnejad and Nicholas Meade and Arkil Patel and Dongchan Shin and Alejandra Zambrano and Karolina Sta\'nczak and Peter Shaw and Christopher J. Pal and Siva Reddy},
  booktitle = {Second Conference on Language Modeling (COLM 2025)},
  year      = {2025},
  url       = {https://colm.cc/virtual/2025/poster/571},
  note      = {Originally released as arXiv:2504.08942}
}

% OLDKEY: arxiv2602-12670 -> NEWKEY: li2026skillsbench
@misc{li2026skillsbench,
      title={{SkillsBench}: Benchmarking How Well Agent Skills Work Across Diverse Tasks},
      author={Xiangyi Li and Yimin Liu and Wenbo Chen and Bingran You and Zonglin Di and Yifeng He and Shenghan Zheng and Kyoung Whan Choe and Jiankai Sun and Shuyi Wang and Chujun Tao and Binxu Li and Xuandong Zhao and Hejia Geng and Xiaojun Wu and Junwei Zhou and Xiaokun Chen and Hanwen Xing and Yubo Li and Qunhong Zeng and Di Wang and Yuanli Wang and Roey Ben Chaim and Penghao Jiang and Haotian Shen and Luyang Kong and Xinyi Liu and Runhui Wang and Xuanqing Liu and Jiachen Li and Xin Lan and Yueqian Lin and Wengao Ye and Junwei He and Songlin Li and Yue Zhang and Yipeng Gao and Yijiang Li and Ze Ma and Liqiang Jing and Tianyu Wang and Kaixin Li and Yiqi Xue and Haoran Lyu and Yizhuo He and Yuchen Tian and Shutong Wu and Bowei Wang and Yixuan Gao and Bo Chen and Litong Liu and Sikai Cheng and Jiajun Bao and Shuaicheng Tong and Shuwen Xu and Terry Yue Zhuo and Tinghan Ye and Qi Qi and Miao Li and Longtai Liao and Zelin Tan and Chang Shi and Xilin Tang and Srinath Tankasala and Boqin Yuan and Yaoyao Qian and Jianhong Tu and Chenguang Wang and Yizhou Sun and Wei Wang and Aaron Taylor and Ziyue Yang and Changkun Guan and Zhikang Dong and Xinyu Zhang and Steven Dillmann and Han-chung Lee and Dawn Song},
      year={2026},
      eprint={2602.12670},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2602.12670}
}

% OLDKEY: arxiv2604-08377 -> NEWKEY: ma2026skillclaw
@misc{ma2026skillclaw,
      title={{SkillClaw}: Let Skills Evolve Collectively with Agentic Evolver},
      author={Ziyu Ma and Shidong Yang and Yuxiang Ji and Xucong Wang and Yong Wang and Yiming Hu and Tongwen Huang and Xiangxiang Chu},
      year={2026},
      eprint={2604.08377},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2604.08377}
}

% OLDKEY: arxiv2605-06614 -> NEWKEY: ouyang2026skillos
@misc{ouyang2026skillos,
      title={{SkillOS}: Learning Skill Curation for Self-Evolving Agents},
      author={Siru Ouyang and Jun Yan and Yanfei Chen and Rujun Han and Zifeng Wang and Bhavana Dalvi Mishra and Rui Meng and Chun-Liang Li and Yizhu Jiao and Kaiwen Zha and Maohao Shen and Vishy Tirumalashetty and George Lee and Jiawei Han and Tomas Pfister and Chen-Yu Lee},
      year={2026},
      eprint={2605.06614},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2605.06614}
}

% OLDKEY: arxiv2608-16393 -> NEWKEY: ying2026deepseek
@misc{ying2026deepseek,
      title={Security Assessment of DeepSeek Harness with A.I.G: Evaluating Resistance to Indirect Prompt Injection},
      author={Zonghao Ying and Xiangfan Wu and Huiyu Wu and Xing Zheng and Huangsheng Cheng and Xiaorong Shi and Jing Guo},
      year={2026},
      eprint={2608.16393},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      note={arXiv preprint},
  url           = {https://arxiv.org/abs/2608.16393}
}


% Boundary case for embodied RSI; model/data update rather than harness-only evolution.
@article{bousmalis2024robocat,
  title   = {{RoboCat}: A Self-Improving Generalist Agent for Robotic Manipulation},
  author  = {Bousmalis, Konstantinos and Vezzani, Giulia and Rao, Dushyant and Devin, Coline and Lee, Alex X. and Bauza, Maria and Davchev, Todor and Zhou, Yuxiang and Gupta, Agrim and Raju, Akhil and Laurens, Antoine and Fantacci, Claudio and Dalibard, Valentin and Zambelli, Martina and Martins, Murilo Fernandes and Pevceviciute, Rugile and Blokzijl, Michiel and Denil, Misha and Batchelor, Nathan and others},
  journal = {Transactions on Machine Learning Research},
  year    = {2024},
  note    = {Originally released as arXiv:2306.11706},
  url     = {https://arxiv.org/abs/2306.11706}
}

% Robot harness self-evolution: fixed external LLM API (base model not trained),
% external multi-agent harness with skill/code refinement; research preview, not peer-reviewed.
@misc{noematrix2026roborsi,
  title        = {{RoboRSI}: Stable, Efficient, and Reusable Robot Self-Evolution in Complex Real-World Environments},
  author       = {{Noematrix Team}},
  year         = {2026},
  month        = sep,
  howpublished = {Research preview; code repository and research blog},
  note         = {\url{https://github.com/nssmd/RoboRSI}; project page \url{https://lab.noematrix.ai/blog/2-roborsi-research-preview/}. Not peer-reviewed}
}

% Epistemic lineage for the practice--cognition--practice organizing lens (Section 2).
% Bibliographic details verified against authoritative sources (2026-09-18).
@incollection{mao1937practice,
  author       = {Mao, Zedong},
  title        = {{On Practice}},
  booktitle    = {Selected Works of Mao Tse-tung},
  volume       = {I},
  publisher    = {Foreign Languages Press},
  address      = {Peking},
  year         = {1965},
  note         = {Essay originally delivered July 1937}
}

@book{popper1959logic,
  author       = {Popper, Karl R.},
  title        = {The Logic of Scientific Discovery},
  publisher    = {Hutchinson},
  address      = {London},
  year         = {1959},
  note         = {English translation and revision of Logik der Forschung, 1935}
}

@book{wiener1948cybernetics,
  author       = {Wiener, Norbert},
  title        = {Cybernetics: Or Control and Communication in the Animal and the Machine},
  publisher    = {The Technology Press and John Wiley \& Sons},
  address      = {Cambridge, MA and New York},
  year         = {1948},
  note         = {Also co-published by Hermann et Cie, Paris}
}

@misc{cagatan2026rewardhacking,
  title        = {Reward Hacking in Language Model Agents: Revisiting {AI} Safety Gridworlds},
  author       = {{\"O}mer Veysel {\c{C}}a{\u{g}}atan and Xuandong Zhao},
  year         = {2026},
  eprint       = {2606.15385},
  archivePrefix= {arXiv},
  primaryClass = {cs.LG},
  note         = {arXiv preprint},
  url           = {https://arxiv.org/abs/2606.15385}
}

@misc{doshi2026verifiably,
  title        = {Towards Verifiably Safe Tool Use for {LLM} Agents},
  author       = {Aarya Doshi and Yining Hong and Congying Xu and Eunsuk Kang and Alexandros Kapravelos and Christian K{\"a}stner},
  year         = {2026},
  eprint       = {2601.08012},
  archivePrefix= {arXiv},
  primaryClass = {cs.SE},
  note         = {arXiv preprint},
  url           = {https://arxiv.org/abs/2601.08012}
}

@misc{jiang2025medagentbench,
  title        = {{MedAgentBench}: A Realistic Virtual {EHR} Environment to Benchmark Medical {LLM} Agents},
  author       = {Yixing Jiang and Kameron C. Black and Gloria Geng and Danny Park and James Zou and Andrew Y. Ng and Jonathan H. Chen},
  year         = {2025},
  eprint        = {2501.14654},
  archivePrefix= {arXiv},
  primaryClass = {cs.AI},
  note         = {arXiv preprint; associated with NEJM AI},
  url           = {https://arxiv.org/abs/2501.14654}
}

@misc{schmidgall2024agentclinic,
  title        = {{AgentClinic}: A Multimodal Agent Benchmark to Evaluate {AI} in Simulated Clinical Environments},
  author       = {Samuel Schmidgall and Rojin Ziaei and Carl Harris and Eduardo Reis and Jeffrey Jopling and Michael Moor},
  year         = {2024},
  eprint        = {2405.07960},
  archivePrefix= {arXiv},
  primaryClass = {cs.AI},
  note         = {arXiv preprint; later in npj Digital Medicine},
  url           = {https://arxiv.org/abs/2405.07960}
}

@misc{lee2026rhi,
  title        = {Recursive Harness Self-Improvement},
  author       = {Hyunin Lee and Jinglue Xu and Jeffrey Seely and Donghyun Lee and Matei Zaharia and Yujin Tang},
  year         = {2026}, eprint = {2607.15524}, archivePrefix = {arXiv}, primaryClass = {cs.AI}, note = {arXiv preprint},
  url = {https://arxiv.org/abs/2607.15524}
}
@misc{chen2026recursivesi,
  title        = {Recursive Self-Improvement in {AI}: From Bounded Self-Refinement to Autonomous Research Loops},
  author       = {Mingguang Chen and Licheng Wang and Bo Qu},
  year         = {2026}, eprint = {2607.07663}, archivePrefix = {arXiv}, primaryClass = {cs.AI}, note = {survey; arXiv preprint},
  url = {https://arxiv.org/abs/2607.07663}
}
@misc{butler1863darwin,
  title        = {Darwin among the Machines},
  author       = {Samuel Butler},
  year         = {1863},
  howpublished = {Letter to the editor (pseud. Cellarius), {\em The Press}, Christchurch, New Zealand},
  note         = {13 June 1863}
}
% (EURISKO already defined above as lenat1983eurisko; duplicate append removed)
@misc{yuan2024selfrewarding,
  title        = {Self-Rewarding Language Models},
  author       = {Weizhe Yuan and Richard Yuanzhe Pang and Kyunghyun Cho and Xian Li and Sainbayar Sukhbaatar and Jing Xu and Jason Weston},
  year         = {2024}, eprint = {2401.10020}, archivePrefix = {arXiv}, primaryClass = {cs.CL}, note = {ICML 2024}
}
@misc{wu2024metarewarding,
  title        = {Meta-Rewarding Language Models: Self-Improving Alignment with {LLM-as-a-Meta-Judge}},
  author       = {Tianhao Wu and Weizhe Yuan and Olga Golovneva and Jing Xu and Yuandong Tian and Jiantao Jiao and Jason Weston and Sainbayar Sukhbaatar},
  year         = {2024}, eprint = {2407.19594}, archivePrefix = {arXiv}, primaryClass = {cs.CL},
  url = {https://arxiv.org/abs/2407.19594}
}

@misc{bai2022constitutional,
  title = {Constitutional {AI}: Harmlessness from {AI} Feedback},
  author = {Yuntao Bai and Saurav Kadavath and Sandipan Kundu and Amanda Askell and Jackson Kernion and Andy Jones and others},
  year = {2022}, eprint = {2212.08073}, archivePrefix = {arXiv}, primaryClass = {cs.CL}, note = {Anthropic; arXiv preprint}
}
@misc{xia2026metaclaw,
  title = {{MetaClaw}: Just Talk---An Agent That Meta-Learns and Evolves in the Wild},
  author = {Peng Xia and Jianwen Chen and Xinyu Yang and Haoqin Tu and Jiaqi Liu and Kaiwen Xiong and others},
  year = {2026}, eprint = {2603.17187}, archivePrefix = {arXiv}, primaryClass = {cs.AI}, note = {arXiv preprint}
}
@misc{lao2026agentx,
  title = {{AgentX}: Towards Agent-Driven Self-Iteration of Industrial Recommender Systems},
  author = {Changxin Lao and others},
  year = {2026}, eprint = {2606.26859}, archivePrefix = {arXiv}, primaryClass = {cs.IR}, note = {Kuaishou; alphabetical authorship; arXiv preprint}
}
@misc{fei2026autoresearch,
  title = {How Do Agents Fail on {AutoResearch}: End-to-End Diagnostic Evaluation on 100 Real-World Frontier Research Tasks},
  author = {Yanlin Fei and Nazhou Liu and Xinmiao Yu and Shaolong Chen and Lei Li and Rahul Thapa and others},
  year = {2026}, eprint = {2608.14905}, archivePrefix = {arXiv}, primaryClass = {cs.AI}, note = {AutoResearchEval; arXiv preprint},
  url = {https://arxiv.org/abs/2608.14905}
}
@article{lenat1984why,
  title = {Why {AM} and {EURISKO} Appear to Work},
  author = {Douglas B. Lenat and John Seely Brown},
  journal = {Artificial Intelligence}, volume = {23}, number = {3}, pages = {269--294}, year = {1984}
}

@misc{jain2024livecodebench,
  author        = {Naman Jain and King Han and Alex Gu and Wen-Ding Li and Fanjia Yan and Tianjun Zhang and Sida Wang and Armando Solar-Lezama and Koushik Sen and Ion Stoica},
  title         = {{LiveCodeBench}: Holistic and Contamination Free Evaluation of Large Language Models for Code},
  year          = {2024},
  eprint        = {2403.07974},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SE},
  url           = {https://arxiv.org/abs/2403.07974}
}

@misc{aleithan2024swebenchplus,
  author        = {Reem Aleithan and Haoran Xue and Mohammad Mahdi Mohajer and Elijah Nnorom and Gias Uddin and Song Wang},
  title         = {{SWE-Bench+}: Enhanced Coding Benchmark for {LLMs}},
  year          = {2024},
  eprint        = {2410.06992},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SE},
  url           = {https://arxiv.org/abs/2410.06992}
}

@misc{zhuge2024agentasjudge,
  author        = {Mingchen Zhuge and Changsheng Zhao and Dylan Ashley and Wenyi Wang and Dmitrii Khizbullin and Yunyang Xiong and Zechun Liu and Ernie Chang and Raghuraman Krishnamoorthi and Yuandong Tian and Yangyang Shi and Vikas Chandra and J{\"u}rgen Schmidhuber},
  title         = {Agent-as-a-Judge: Evaluate Agents with Agents},
  year          = {2024},
  eprint        = {2410.10934},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2410.10934}
}

@misc{yu2025polyskill,
  author        = {Simon Yu and Gang Li and Weiyan Shi and Peng Qi},
  title         = {{PolySkill}: Learning Generalizable Skills Through Polymorphic Abstraction},
  year          = {2025},
  eprint        = {2510.15863},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2510.15863}
}

@misc{wang2025huxley,
  author        = {Wenyi Wang and Piotr Pi{\k{e}}kos and Li Nanbo and Firas Laakom and Yimeng Chen and Mateusz Ostaszewski and Mingchen Zhuge and J{\"u}rgen Schmidhuber},
  title         = {Huxley-{G{\"o}del} Machine: Human-Level Coding Agent Development by an Approximation of the Optimal Self-Improving Machine},
  year          = {2025},
  eprint        = {2510.21614},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2510.21614}
}

@misc{xia2025liveswe,
  author        = {Chunqiu Steven Xia and Zhe Wang and Yan Yang and Yuxiang Wei and Lingming Zhang},
  title         = {Live-{SWE}-agent: Can Software Engineering Agents Self-Evolve on the Fly?},
  year          = {2025},
  eprint        = {2511.13646},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SE},
  url           = {https://arxiv.org/abs/2511.13646}
}

@misc{chen2026harnessfix,
  author        = {Mengzhuo Chen and Junjie Wang and Zhe Liu and Yawen Wang and Haiming Zheng and Qing Wang},
  title         = {From Failed Trajectories to Reliable {LLM} Agents: Diagnosing and Repairing Harness Flaws},
  year          = {2026},
  eprint        = {2606.06324},
  archivePrefix = {arXiv},
  primaryClass  = {cs.SE},
  url           = {https://arxiv.org/abs/2606.06324}
}

@misc{luo2026harnessbank,
  author        = {Xiaotian Luo and Dizhan Xue and Fengxingyu Wang and Chuanrui Hu and Yafeng Deng},
  title         = {{HarnessBank}: Semantic Gene-Bank Search with Gated Verification for Agent-Harness Self-Evolution},
  year          = {2026},
  eprint        = {2607.13683},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2607.13683}
}

@misc{zhang2026harnesscompass,
  author        = {Luan Zhang and Ruochen Zhou and Dandan Song and Zhengyu Chen and Yuhang Tian and Jun Yang and Huipeng Ma and Chenhao Li and Guangyuan Feng and Xudong Li and Yizhou Jin and Yan Xu},
  title         = {{HarnessCompass}: Guiding Automatic Harness Evolution toward Generalizable and Effective Agent Harnesses},
  year          = {2026},
  eprint        = {2608.01918},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2608.01918}
}

@article{guha2023legalbench,
  title={{LegalBench}: A collaboratively built benchmark for measuring legal reasoning in large language models},
  author={Guha, Neel and Nyarko, Julian and Ho, Daniel and R{\'e}, Christopher and Chilton, Adam and K, Aditya and Chohlas-Wood, Alex and Peters, Austin and Waldon, Brandon and Rockmore, Daniel and Zambrano, Diego and others},
  journal={Advances in Neural Information Processing Systems},
  volume={36},
  pages={44123--44279},
  year={2023}
}

@inproceedings{mozannar2024reading,
  title={Reading between the lines: Modeling user behavior and costs in {AI}-assisted programming},
  author={Mozannar, Hussein and Bansal, Gagan and Fourney, Adam and Horvitz, Eric},
  booktitle={Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems},
  pages={1--16},
  year={2024}
}

@article{peng2023impact,
  title={The impact of {AI} on developer productivity: Evidence from {GitHub Copilot}},
  author={Peng, Sida and Kalliamvakou, Eirini and Cihon, Peter and Demirer, Mert},
  journal={arXiv preprint arXiv:2302.06590},
  year={2023}
}

@article{forsgren2021space,
  title={The {SPACE} of Developer Productivity: There's more to it than you think.},
  author={Forsgren, Nicole and Storey, Margaret-Anne and Maddila, Chandra and Zimmermann, Thomas and Houck, Brian and Butler, Jenna},
  journal={Queue},
  volume={19},
  number={1},
  pages={20--48},
  year={2021},
  publisher={ACM New York, NY, USA}
}

@article{liubeyond2026,
author = {Liu, Fei and Mckinnon, Todd},
title = {Beyond the Hype: The Efficiency-Throughput Gap with {GitHub} {Copilot}},
year = {2026},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
volume = {69},
number = {10},
issn = {0001-0782},
journal = {Commun. ACM},
pages = {68--77},
doi = {10.1145/3797488}
}

@article{acemoglu2018race,
  title={The race between man and machine: Implications of technology for growth, factor shares, and employment},
  author={Acemoglu, Daron and Restrepo, Pascual},
  journal={American Economic Review},
  volume={108},
  number={6},
  pages={1488--1542},
  year={2018},
  publisher={American Economic Association 2014 Broadway, Suite 305, Nashville, TN 37203}
}

@article{citron2007technological,
  title={Technological due process},
  author={Citron, Danielle Keats},
  journal={Wash. UL Rev.},
  volume={85},
  pages={1249},
  year={2008},
  publisher={HeinOnline}
}

%% ==========================================================================
% ADDED 2026-09-29. Entries adopted after a survey sweep of the two
% neighbouring self-evolution surveys (arXiv:2607.13104, arXiv:2608.03392).
% Verification for this block: all eprint IDs were resolved against the
% arXiv API (title, first author and v1 date compared field by field); all
% DOIs against Crossref; all documentation URLs by following redirects to a
% live final URL. Where a source bibliography claimed a venue that could
% only be checked on OpenReview (unreachable, bot challenge), the entry is
% recorded as a preprint rather than upgraded on secondary evidence.
%% ==========================================================================


% --- Benchmarks built to evaluate self-evolution (Section 5.3) ---

@misc{zhang2026evocodebench,
  title         = {{EvoCodeBench}: A Human-Performance Benchmark for Self-Evolving {LLM}-Driven Coding Systems},
  author        = {Zhang, Wentao and Wang, Jianfeng and Liang, Liheng and Zhao, Yilei and Wen, Haibin and Zhao, Zhe},
  year          = {2026},
  eprint        = {2602.10171},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2602.10171}
}

@misc{le2025sweevo,
  title         = {{SWE}-{EVO}: Benchmarking Coding Agents in Long-Horizon Software Evolution Scenarios},
  author        = {Le, Tue and Thai, Minh V. T. and Manh, Dung Nguyen and Nhat, Huy Phan and Bui, Nghi D. Q.},
  year          = {2025},
  eprint        = {2512.18470},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2512.18470}
}

@misc{zheng2025lifelongagentbench,
  title         = {{LifelongAgentBench}: Evaluating {LLM} Agents as Lifelong Learners},
  author        = {Zheng, Junhao and Cai, Xidi and Li, Qiuke and Zhang, Duzhen and Li, Zhongzhi and Zhang, Yingying and Song, Le and Ma, Qianli},
  year          = {2025},
  eprint        = {2505.11942},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2505.11942}
}

@misc{zhang2026skillflow,
  title         = {{SkillFlow}: Benchmarking Lifelong Skill Discovery and Evolution for Autonomous Agents},
  author        = {Zhang, Ziao and Shi, Kou and Huang, Shiting and Nie, Avery and Zeng, Yu and Zhao, Yiming and Fang, Zhen and Su, Qishen and Qiu, Haibo and Yang, Wei and Ren, Qingnan and Zou, Shun and others},
  year          = {2026},
  eprint        = {2604.17308},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.17308}
}

@misc{zhong2026skilllearnbench,
  title         = {{SkillLearnBench}: Benchmarking Continual Learning Methods for Agent Skill Generation on Real-World Tasks},
  author        = {Zhong, Shanshan and Lu, Yi and Ning, Jingjie and Wan, Yibing and Feng, Lihan and Ao, Yuyi and Ribeiro, Leonardo F. R. and Dreyer, Markus and Ammirati, Sean and Xiong, Chenyan},
  year          = {2026},
  eprint        = {2604.20087},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.20087}
}

@misc{xue2026pastbench,
  title         = {{PAST}-{Bench}: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents},
  author        = {Xue, Shuhan and Ding, Zixin and Shen, Yichen and Wang, Yinjie and Yin, Zhenfei and Wu, Yingcheng and Chen, Yuxin and Wang, Mengdi and Yang, Ling},
  year          = {2026},
  eprint        = {2608.04003},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.04003}
}

@misc{jiang2026seaeval,
  title         = {{SEA}-{Eval}: A Benchmark for Evaluating Self-Evolving Agents Beyond Episodic Assessment},
  author        = {Jiang, Sihang and Ma, Lipeng and Hong, Zhonghua and Wang, Keyi and Lu, Zhiyu and Wang, Tengfei and Chen, Shisong and Zhang, Jinghao and Pan, Tianjun and Li, Weijia and Liang, Jiaqing and Xiao, Yanghua},
  year          = {2026},
  eprint        = {2604.08988},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.08988}
}

@misc{he2026memoryarena,
  title         = {{MemoryArena}: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks},
  author        = {He, Zexue and Wang, Yu and Zhi, Churan and Hu, Yuanzhe and Chen, Tzu-Ping and Yin, Lang and Chen, Ze and Wu, Tong Arthur and Ouyang, Siru and Wang, Zihan and Pei, Jiaxin and McAuley, Julian and Choi, Yejin and Pentland, Alex},
  year          = {2026},
  eprint        = {2602.16313},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2602.16313}
}

@misc{wei2025evomemory,
  title         = {{Evo}-{Memory}: Benchmarking {LLM} Agent Test-time Learning with Self-Evolving Memory},
  author        = {Wei, Tianxin and Sachdeva, Noveen and Coleman, Benjamin and He, Zhankui and Bei, Yuanchen and Ning, Xuying and Ai, Mengting and Li, Yunzhe and He, Jingrui and Chi, Ed H. and Wang, Chi and Chen, Shuo and Pereira, Fernando and Kang, Wang-Cheng and Cheng, Derek Zhiyuan},
  year          = {2025},
  eprint        = {2511.20857},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2511.20857}
}

@misc{zhou2026gdpevo,
  title         = {{GDPevo}: Evaluating Agent Self-Evolution on Real Business Tasks},
  author        = {Zhou, Leijun and Liu, Zhihao and Qu, Xiang and Liu, Chenxu and Liu, Yifei and Yu, Yanke and Xu, Jingzhe and Wu, Xuejun and Qian, Buyue and Chen, Xi and Zheng, Yaowei and Hu, Junhao},
  year          = {2026},
  eprint        = {2608.03764},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.03764}
}

@misc{deng2026finevobench,
  title         = {{FinEvo}-{Bench}: A Longitudinal Benchmark for Self-Evolving Agents in Professional Financial Workflows},
  author        = {Deng, Bo and Zhou, Kang and Guo, Lifan and Tao, Chongyang and Chen, Xuanren and Xie, Chenggang and Liang, Renzhao and Chen, Feng and Zhang, Chi},
  year          = {2026},
  eprint        = {2608.06144},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.06144}
}

@misc{guan2026continualskillbench,
  title         = {{ContinualSkillBench}: Can {LLM} Agents Truly Evolve Their Capabilities?},
  author        = {Guan, Tianyi and Wang, Yiding and Yang, Haotong and Cao, Siyuan and Liu, Shirui and Hu, Yi and Li, Jiaqi and Zhang, Muhan},
  year          = {2026},
  eprint        = {2608.03874},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.03874}
}

@misc{asawa2026clbench,
  title         = {Continual Learning Bench: Evaluating Frontier {AI} Systems in Real-World Stateful Environments},
  author        = {Asawa, Parth and Glaze, Christopher M. and Orlanski, Gabriel and Ramakrishnan, Ramya and Xu, Benji and Biswal, Asim and Chen, Vincent Sunn and Sala, Frederic and Zaharia, Matei and Gonzalez, Joseph E.},
  year          = {2026},
  eprint        = {2606.05661},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.05661}
}

@misc{xia2026toolgenesis,
  title         = {{Tool}-{Genesis}: A Task-Driven Tool Creation Benchmark for Self-Evolving Language Agent},
  author        = {Xia, Bowei and Hu, Mengkang and Wang, Shijian and Jin, Jiarui and Jiao, Wenxiang and Lu, Yuan and Li, Kexin and Luo, Ping},
  year          = {2026},
  eprint        = {2603.05578},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2603.05578}
}

@misc{bai2026agentcost,
  title         = {How Do {AI} Agents Spend Your Money? Analyzing and Predicting Token Consumption in Agentic Coding Tasks},
  author        = {Bai, Longju and Huang, Zhemin and Wang, Xingyao and Sun, Jiao and Mihalcea, Rada and Brynjolfsson, Erik and Pentland, Alex and Pei, Jiaxin},
  year          = {2026},
  eprint        = {2604.22750},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.22750}
}


% --- Contamination-controlled coding suites (Section 5.3.2) ---

@misc{badertdinov2025swerebench,
  title         = {{SWE}-rebench: An Automated Pipeline for Task Collection and Decontaminated Evaluation of Software Engineering Agents},
  author        = {Badertdinov, Ibragim and Golubev, Alexander and Nekrashevich, Maksim and Shevtsov, Anton and Karasik, Simon and Andriushchenko, Andrei and Trofimova, Maria and Litvintseva, Daria and Yangel, Boris},
  year          = {2025},
  eprint        = {2505.20411},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2505.20411}
}

@misc{zhang2025swebenchlive,
  title         = {{SWE}-bench Goes Live!},
  author        = {Zhang, Linghao and He, Shilin and Zhang, Chaoyun and Kang, Yu and Li, Bowen and Xie, Chengxing and Wang, Junhao and Wang, Maoquan and Huang, Yufan and Fu, Shengyu and Nallipogu, Elsie and Lin, Qingwei and Dang, Yingnong and Rajmohan, Saravan and Zhang, Dongmei},
  year          = {2025},
  eprint        = {2505.23419},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2505.23419}
}


% --- Harness-attribution critique line (Sections 1 and 5) ---

@misc{bensghaier2026blame,
  title         = {Don't Blame the Large Language Model: How Agent Harness Evolution Shapes Coding Agent Quality},
  author        = {Ben Sghaier, Oussama and Li, Hao and Adams, Bram and Hassan, Ahmed E.},
  year          = {2026},
  eprint        = {2607.03691},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.03691}
}

@misc{vats2026scaffold,
  title         = {The Scaffold Effect in Coding Agents: Harness Choice as a Hidden Variable in Coding-Agent Evaluation},
  author        = {Vats, Naman and Golev, Oleg},
  year          = {2026},
  eprint        = {2607.22585},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.22585}
}

@misc{wang2026rethinking,
  title         = {Rethinking the Evaluation of Harness Evolution for Agents},
  author        = {Wang, Yike and Zhu, Huaisheng and Hu, Zhengyu and Yuan, Yige and Chen, Zhengyu and Senthil, Shakti and Hajishirzi, Hannaneh and Tsvetkov, Yulia and Smith, Noah A. and Dasigi, Pradeep and Xiao, Teng},
  year          = {2026},
  eprint        = {2607.12227},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.12227}
}

@misc{wang2026tampering,
  title         = {Auditing Harness Tampering in Self-Improving Agents},
  author        = {Wang, Xing and Zhang, Xiaoyi and Shao, Jie},
  year          = {2026},
  eprint        = {2609.00069},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2609.00069}
}

@misc{wang2026phantom,
  title         = {{Phantom} {Guardrails}: When Self-Improving Agent Harnesses Fix Failures That Never Happened},
  author        = {Wang, Su and Qian, Pin and Lin, Yifan and Xu, Jingzhou and Chen, Yihang and Jiang, Xiaochong and Liu, Lifei and Yu, Haoran},
  year          = {2026},
  eprint        = {2607.13083},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.13083}
}

@misc{guo2026selfverification,
  title         = {Self-Authored Verification Is Unreliable in Heuristic Self-Improving Agents},
  author        = {Guo, Diandian and Cao, Cong and Yuan, Fangfang and Wang, Yingqi and Wang, Yueshan and Wang, Dakui},
  year          = {2026},
  eprint        = {2607.24300},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.24300}
}


% --- Admission and retirement gates for accumulated state (Section 5.2) ---

@misc{zhang2026ratchet,
  title         = {{Ratchet}: How Reliable Must an {LLM} Judge Be to Retire a Skill?},
  author        = {Xing Zhang and Yanwei Cui and Guanghui Wang and Ziyuan Li and Wei Qiu and Bing Zhu and Peiyang He},
  year          = {2026},
  eprint        = {2605.22148},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.22148}
}

@misc{liu2026skillsvote,
  title         = {{SkillsVote}: Lifecycle Governance of Agent Skills from Collection, Recommendation to Evolution},
  author        = {Hongyi Liu and Haoyan Yang and Tao Jiang and Bo Tang and Feiyu Xiong and Yuyu Luo and Zhiyu Li},
  year          = {2026},
  eprint        = {2605.18401},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.18401}
}

@misc{zhang2026gradergrader,
  title         = {Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving {LLM} Agents},
  author        = {Xing Zhang and Guanghui Wang and Yanwei Cui and Ziyuan Li and Wei Qiu and Bing Zhu and Peiyang He},
  year          = {2026},
  eprint        = {2607.12790},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.12790}
}

@misc{zhang2026librarydrift,
  title         = {{Library} {Drift}: Diagnosing and Fixing a Silent Failure Mode in Self-Evolving {LLM} Skill Libraries},
  author        = {Zhang, Xing and Cui, Yanwei and Wang, Guanghui and Li, Ziyuan and Qiu, Wei and Zhu, Bing and He, Peiyang},
  year          = {2026},
  eprint        = {2605.19576},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.19576}
}

@misc{shang2026backfires,
  title         = {When Self-Evolution Backfires: Pre-Commit Gating against Skill Contamination in {LLM} Agents},
  author        = {Linfang Shang and Ming Xu and Yiding Sun and Tianle Xia and Lingxiang Hu and Lan Xu and Ning Zheng},
  year          = {2026},
  eprint        = {2608.05810},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.05810}
}

@misc{wu2026evomal,
  title         = {{EVOMAL}: Self-Poisoning in Self-Evolving Coding Agents},
  author        = {Xiaodong Wu and Yu Shi and Qi Li and Zhimin Zhao and Xiangman Li and Bram Adams and Ahmed E. Hassan and Jianbing Ni},
  year          = {2026},
  eprint        = {2608.25776},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.25776}
}

@misc{asadolahi2026inflation,
  title         = {Memory Reward Inflation in Self-Improving {LLM} Agents},
  author        = {Asadolahi, Mohammad and Amini, Amir and Talebi, Samira and Farhadi, Amirfarhad and Zamanifar, Azadeh},
  year          = {2026},
  eprint        = {2608.00017},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.00017}
}


% --- Executable harness update targets (Section 4) ---

@misc{park2026autosaddler,
  title         = {{AutoSaddler}: Automatic Harness Optimization with Durable Updates from Agent Execution Traces},
  author        = {Sungho Park and Wonjoong Kim and Rongyuan Tan and Jue Zhang and Wook-Shin Han and Pengfei Gao and Chanyoung Park and Yongqiang Yao and Rao Fu and Elsie Nallipogu and Qingwei Lin and Saravan Rajmohan and Dongmei Zhang},
  year          = {2026},
  eprint        = {2608.23041},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.23041}
}

@misc{wei2026evoharness,
  title         = {{Evo}-{Harness}: Context-to-Harness Skill Compilation for Self-Evolving Agents},
  author        = {Tianxin Wei and Zhan Shi and Minhua Lin and Bing He and Zewen Liu and Yisi Sang and Yuanchen Bei and Xuying Ning and Jiaru Zou and Ting-Wei Li and Xiao Lin and Yanjun Zhao and Chi Wang and Benoit Dumoulin and Dakuo Wang and Jingrui He and Hanqing Lu},
  year          = {2026},
  eprint        = {2608.15071},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.15071}
}

@misc{hao2026recreate,
  title        = {{ReCreate}: Reasoning and Creating Domain Agents Driven by Experience},
  author       = {Zhezheng Hao and Hong Wang and Jian Luo and Jianqing Zhang and Yuyan Zhou and Qiang Lin and Can Wang and Hande Dong and Jiawei Chen},
  year         = {2026},
  howpublished = {\url{https://aclanthology.org/2026.acl-long.1432/}}
}

@misc{yan2026harnessofharness,
  title         = {{Harness}-of-{Harness}: Multi-Day Autonomous Software Development with Continual Improvement},
  author        = {Haoyang Yan and Min-le Su and Hangfan Zhang and Zhanhao Li and Chen Zhang and Shao Zhang and Yang Chen and Lei Bai and Shuyue Hu},
  year          = {2026},
  eprint        = {2609.01481},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2609.01481}
}

@misc{yang2026onerecipe,
  title         = {One Recipe, Many Harnesses: What Self-Evolution Encodes Across Languages and Models},
  author        = {Siqi Yang and Qianlan Yang and Yu-Xiong Wang and Saurabh Pujar and Martin Hirzel},
  year          = {2026},
  eprint        = {2608.10178},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.10178}
}

@misc{nie2026evolvenet,
  title         = {{EvolveNet}: Collaborative Harness Evolution for Agent Self-Improvement},
  author        = {Jun Nie and Yonggang Zhang and Qianshu Cai and Yiu-ming Cheung and Xinmei Tian and Bo Han},
  year          = {2026},
  eprint        = {2608.04968},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.04968}
}

@misc{zhang2026autogenesis,
  title         = {{Autogenesis}: A Self-Evolving Agent Protocol},
  author        = {Wentao Zhang and Zhe Zhao and Haibin Wen and Yingcheng Wu and Cankun Guo and Ming Yin and Bo An},
  year          = {2026},
  eprint        = {2604.15034},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.15034}
}

@misc{nie2026tthe,
  title         = {{TTHE}: Test-Time Harness Evolution},
  author        = {Jun Nie and Yonggang Zhang and Jun Song and Qianshu Cai and Dahai Yu and Yike Guo and Xinmei Tian and Bo Han},
  year          = {2026},
  eprint        = {2607.08124},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.08124}
}

@misc{xu2026lifeharness,
  title         = {Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic {LLM} Agents},
  author        = {Tianshi Xu and Huifeng Wen and Meng Li},
  year          = {2026},
  eprint        = {2605.22166},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.22166}
}

@misc{liu2026autoharness,
  title         = {{Adaptive} {Auto}-{Harness}: Sustained Self-Improvement for Agentic System Deployment on Open-Ended Task Streams},
  author        = {Zewen Liu and Zhan Shi and Yisi Sang and Bing He and Minhua Lin and Tianxin Wei and Dakuo Wang and Benoit Dumoulin and Wei Jin and Hanqing Lu},
  year          = {2026},
  eprint        = {2606.01770},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.01770}
}

@misc{zhang2026agentdevel,
  title         = {{AgentDevel}: Reframing Self-Evolving {LLM} Agents as Release Engineering},
  author        = {Di Zhang},
  year          = {2026},
  eprint        = {2601.04620},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2601.04620}
}

@misc{nguyen2026rsea,
  title         = {Recursive Self-Evolving Agents via Held-Out Selection},
  author        = {Michael Nguyen and Quoc Nguyen and Paul Vuong},
  year          = {2026},
  eprint        = {2606.28374},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.28374}
}

@misc{borthwick2026robophd,
  title         = {{RoboPhD}: Self-Improving Text-to-{SQL} Through Autonomous Agent Evolution},
  author        = {Andrew Borthwick and Stephen Ash},
  year          = {2026},
  eprint        = {2601.01126},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2601.01126}
}

@misc{iacob2026redqueen,
  title         = {The Red Queen {G{\"o}del} Machine: Co-Evolving Agents and Their Evaluators},
  author        = {Alex Iacob and Andrej Jovanovi{\'c} and William F. Shen and Daniel Burkhardt and Meghdad Kurmanji and Nurbek Tastan and Lorenzo Sani and Niccol{\`o} Alberto Elia Venanzi and Jiayi Nie and Ambroise Odonnat and Zeyu Cao and Bill Marino and Xinchi Qiu and Rika Antonova and Nicholas D. Lane},
  year          = {2026},
  eprint        = {2606.26294},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.26294}
}


% --- Sequential-validity guarantees (Section 6) ---

@misc{sengupta2026sea,
  title         = {Self-Evolving Agents with Anytime-Valid Certificates},
  author        = {Biswa Sengupta},
  year          = {2026},
  eprint        = {2607.00871},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.00871}
}


% --- Workflow and topology update targets (Section 3.3) ---

@misc{ma2026judgeflow,
  title         = {{JudgeFlow}: Agentic Workflow Optimization via Block Judge},
  author        = {Zihan Ma and Zhikai Zhao and Chuanbo Hua and Federico Berto and Jinkyoo Park},
  year          = {2026},
  eprint        = {2601.07477},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2601.07477}
}

@misc{liu2025sew,
  title         = {{SEW}: Self-Evolving Agentic Workflows for Automated Code Generation},
  author        = {Liu, Siwei and Fang, Jinyuan and Zhou, Han and Wang, Yingxu and Meng, Zaiqiao},
  year          = {2025},
  eprint        = {2505.18646},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2505.18646}
}

@misc{zhang2025evoflow,
  title         = {{EvoFlow}: Evolving Diverse Agentic Workflows On The Fly},
  author        = {Guibin Zhang and Kaijie Chen and Guancheng Wan and Heng Chang and Hong Cheng and Kun Wang and Shuyue Hu and Lei Bai},
  year          = {2025},
  eprint        = {2502.07373},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2502.07373}
}

@misc{wang2025evoagentx,
  title         = {{EvoAgentX}: An Automated Framework for Evolving Agentic Workflows},
  author        = {Wang, Yingxu and Liu, Siwei and Fang, Jinyuan and Meng, Zaiqiao},
  year          = {2025},
  eprint        = {2507.03616},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2507.03616}
}

@misc{ren2026flowevo,
  title         = {{FlowEvo}: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills},
  author        = {Zeyu Ren and Ling Yue and Ran Li and Yishu Wang and Shengxiang Xu and Hanmo Liu and Shaowu Pan and Shimin Di},
  year          = {2026},
  eprint        = {2607.21596},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.21596}
}

@misc{weng2026gea,
  title         = {{Group}-{Evolving} {Agents}: Open-Ended Self-Improvement via Experience Sharing},
  author        = {Weng, Zhaotian and Antoniades, Antonis and Nathani, Deepak and Zhang, Zhen and Pu, Xiao and Wang, Xin Eric},
  year          = {2026},
  eprint        = {2602.04837},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2602.04837}
}

@misc{hao2026evolveteam,
  title         = {Evolve as a Team: Collaborative Self-Evolution for {LLM}-based Multi-Agent Systems},
  author        = {Zhezheng Hao and Tianfu Wang and Huanshuo Dong and Ziyan Liu and Hong Wang and Xiankun Lin and Qiang Lin and Can Wang and Hande Dong and Jiawei Chen},
  year          = {2026},
  eprint        = {2605.29790},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.29790}
}

@misc{ye2026evomap,
  title         = {{Behind} {EvoMap}: Characterizing a Self-Evolving Agent-to-Agent Collaboration Network},
  author        = {Ye, Qiming and Zhang, Peixian and He, Yupeng and Peng, Zifan and Tyson, Gareth},
  year          = {2026},
  eprint        = {2605.25815},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.25815}
}


% --- Reusable skill documents (Sections 3.2 and 4.1) ---

@misc{li2026codeskill,
  title         = {{CODESKILL}: Learning Self-Evolving Skills for Coding Agents},
  author        = {Li, Yanzhou and Zhang, Yiran and Zhang, Xiaoyu and Liu, Xiaoxia and Liu, Yang},
  year          = {2026},
  eprint        = {2605.25430},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.25430}
}

@inproceedings{tan2026gskill,
  title     = {Automatically Learning Skills for Coding Agents},
  author    = {Tan, Shangyin and Agrawal, Lakshya A. and Sandadi, Rohit and Klein, Dan and Sen, Koushik and Dimakis, Alexandros G. and Zaharia, Matei},
  year      = {2026},
  booktitle = {Proceedings of the ACM Conference on AI and Agentic Systems},
  doi       = {10.1145/3786335.3813196},
  url       = {https://doi.org/10.1145/3786335.3813196}
}

@misc{xiao2026socraticswe,
  title         = {{Socratic}-{SWE}: Self-Evolving Coding Agents via Trace-Derived Agent Skills},
  author        = {Xiao, Chuan and Jiao, Zhengbo and Wang, Shaobo and Wang, Wei and Zhao, Bing and Wei, Hu and Zhang, Linfeng and Qu, Lin},
  year          = {2026},
  eprint        = {2606.07412},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.07412}
}

@misc{yang2026gse,
  title         = {Learning Globally Reusable Skills for Coding Agents},
  author        = {Chen Yang and Jiashuo Tian and Ziqi Wang and Xinyin Liu and Meiru Ye and Junjie Chen},
  year          = {2026},
  eprint        = {2608.06153},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.06153}
}

@misc{wang2026effiskill,
  title         = {{EffiSkill}: Agent Skill Based Automated Code Efficiency Optimization},
  author        = {Wang, Zimu and Shi, Yuling and Li, Mengfan and Liu, Zijun and Zhang, Jie M. and Wan, Chengcheng and Gu, Xiaodong},
  year          = {2026},
  eprint        = {2603.27850},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2603.27850}
}

@misc{gong2026skillmoo,
  title         = {{SkillMOO}: Multi-Objective Optimization of Agent Skills for Software Engineering},
  author        = {Jingzhi Gong and Ruizhen Gu and Zhiwei Fei and Yazhuo Cao and Lukas Twist and Alina Geiger and Shuo Han and Dominik Sobania and Federica Sarro and Jie M. Zhang},
  year          = {2026},
  eprint        = {2604.09297},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.09297}
}

@misc{huang2026consumption,
  title         = {From Raw Experience to Skill Consumption: A Systematic Study of Model-Generated Agent Skills},
  author        = {Huang, Zisu and Xu, Jingwen and Yang, Yifan and Gong, Ziyang and Yang, Qihao and Tian, Muzhao and Wang, Xiaohua and Lv, Changze and Gao, Xuemei and Dai, Qi and Liu, Bei and Qiu, Kai and Yang, Xue and Chen, Dongdong and Zheng, Xiaoqing and Luo, Chong},
  year          = {2026},
  eprint        = {2605.23899},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.23899}
}

@misc{lin2026muse,
  title         = {{MUSE}-{Autoskill}: Self-Evolving Agents via Skill Creation, Memory, Management, and Evaluation},
  author        = {Lin, Huawei and Li, Peng and Song, Jie and Jiang, Fuxin and Zhang, Tieying},
  year          = {2026},
  eprint        = {2605.27366},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.27366}
}


% --- Persistent context and memory (Section 3.2) ---

@misc{chen2025sweexp,
  title         = {{SWE}-{Exp}: Experience-Driven Software Issue Resolution},
  author        = {Chen, Silin and Lin, Shaoxin and Shi, Yuling and Lian, Heng and Gu, Xiaodong and Yun, Longfei and Chen, Dong and Cao, Lin and Liu, Jiyang and Xia, Nu and Wang, Qianxiang},
  year          = {2025},
  eprint        = {2507.23361},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2507.23361}
}

@misc{tang2025agentkb,
  title         = {{Agent} {KB}: Leveraging Cross-Domain Experience for Agentic Problem Solving},
  author        = {Tang, Xiangru and Qin, Tianrui and Peng, Tianhao and Zhou, Ziyang and Shao, Daniel and Du, Tingting and Wei, Xinming and Xia, Peng and Wu, Fang and Zhu, He and Zhang, Ge and Liu, Jiaheng and Wang, Xingyao and Hong, Sirui and Wu, Chenglin and Cheng, Hao and Wang, Chi and Zhou, Wangchunshu},
  year          = {2025},
  eprint        = {2507.06229},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2507.06229}
}

@misc{gao2026swemem,
  title         = {{SWE}-{MeM}: Learning Adaptive Memory Management for Long-Horizon Coding Agents},
  author        = {Shuzheng Gao and Wenhao Zeng and Zhaojian Yu and Jianqiao Wangni and Chaozheng Wang and Kai Cai and Shilin He and Michael R. Lyu},
  year          = {2026},
  eprint        = {2606.28434},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.28434}
}

@misc{deng2026memcoder,
  title         = {Your Code Agent Can Grow Alongside You with Structured Memory},
  author        = {Yi-Xuan Deng and Xiaoqin Liu and Yi Zhang and Guo-Wei Yang and Shuojin Yang},
  year          = {2026},
  eprint        = {2603.13258},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2603.13258}
}

@misc{aggarwal2026behavioral,
  title         = {Self-Improving {AI} Coding Agents Through Accumulated Behavioral Rules: A Closed-Loop Framework},
  author        = {Aditya Aggarwal and Nahid Farhady Ghalaty},
  year          = {2026},
  eprint        = {2607.13091},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2607.13091}
}

@misc{ren2026taco,
  title         = {A Self-Evolving Framework for Efficient Terminal Agents via Observational Context Compression},
  author        = {Jincheng Ren and Siwei Wu and Yizhi Li and Kang Zhu and Shu Xu and Boyu Feng and Ruibin Yuan and Wei Zhang and Riza Batista-Navarro and Jian Yang and Chenghua Lin},
  year          = {2026},
  eprint        = {2604.19572},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.19572}
}

@misc{wang2026swepruner,
  title         = {{SWE}-{Pruner}: Self-Adaptive Context Pruning for Coding Agents},
  author        = {Yuhang Wang and Yuling Shi and Mo Yang and Rongrui Zhang and Shilin He and Heng Lian and Yuting Chen and Siyu Ye and Kai Cai and Xiaodong Gu},
  year          = {2026},
  eprint        = {2601.16746},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2601.16746}
}

@misc{zhang2025memevolve,
  title         = {{MemEvolve}: Meta-Evolution of Agent Memory Systems},
  author        = {Zhang, Guibin and Ren, Haotian and Zhan, Chong and Zhou, Zhenhong and Wang, Junhao and Zhu, He and Zhou, Wangchunshu and Yan, Shuicheng},
  year          = {2025},
  eprint        = {2512.18746},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2512.18746}
}

@inproceedings{xiong2026memorymgmt,
  title     = {How Memory Management Impacts {LLM} Agents: An Empirical Study of Experience-Following Behavior},
  author    = {Xiong, Zidi and Lin, Yuping and Xie, Wenya and He, Pengfei and Liu, Zirui and Tang, Jiliang and Lakkaraju, Himabindu and Xiang, Zhen},
  year      = {2026},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  doi       = {10.18653/v1/2026.acl-long.27},
  url       = {https://aclanthology.org/2026.acl-long.27/}
}

@misc{kim2026memorytransfer,
  title         = {Memory Transfer Learning: How Memories are Transferred Across Domains in Coding Agents},
  author        = {Kim, Kangsan and Kang, Minki and Kim, Taeil and Yang, Yanlai and Ren, Mengye and Hwang, Sung Ju},
  year          = {2026},
  eprint        = {2604.14004},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.14004}
}


% --- Instruction and prompt update targets (Section 3.1) ---

@misc{cheng2024trace,
  title         = {Trace is the Next {AutoDiff}: Generative Optimization with Rich Feedback, Execution Traces, and {LLMs}},
  author        = {Ching-An Cheng and Allen Nie and Adith Swaminathan},
  year          = {2024},
  eprint        = {2406.16218},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2406.16218},
  note          = {NeurIPS 2024}
}

@misc{xu2025metatextgrad,
  title         = {{metaTextGrad}: Automatically optimizing language model optimizers},
  author        = {Guowei Xu and Mert Yuksekgonul and Carlos Guestrin and James Zou},
  year          = {2025},
  eprint        = {2505.18524},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2505.18524}
}

@misc{tao2025delvepo,
  title         = {{DelvePO}: Direction-Guided Self-Evolving Framework for Flexible Prompt Optimization},
  author        = {Tao Tao and Guanghui Zhu and Lang Guo and Hongyi Chen and Chunfeng Yuan and Yihua Huang},
  year          = {2025},
  eprint        = {2510.18257},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2510.18257}
}

@misc{zhao2025egoprompt,
  title         = {How to Auto-optimize Prompts for Domain Tasks? Adaptive Prompting and Reasoning through Evolutionary Domain Knowledge Adaptation},
  author        = {Yang Zhao and Pu Wang and Hao Frank Yang},
  year          = {2025},
  eprint        = {2510.21148},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2510.21148}
}

@misc{nair2025deevo,
  title         = {Tournament of Prompts: Evolving {LLM} Instructions Through Structured Debates and {Elo} Ratings},
  author        = {Anirudh Nair and Adi Banerjee and Laurent Mombaerts and Matthew Hagen and Tarik Borogovac},
  year          = {2025},
  eprint        = {2506.00178},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2506.00178}
}

@misc{yang2026vaso,
  title         = {{VASO}: Formally Verifiable Self-Evolving Skills for Physical {AI} Agents},
  author        = {Yunhao Yang and Neel P. Bhatt and Kevin Wang and Samuel Tetteh and Zhangyang Wang and Ufuk Topcu},
  year          = {2026},
  eprint        = {2606.05395},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.05395}
}

@misc{tao2026sepo,
  title         = {{SePO}: Self-Evolving Prompt Agent for System Prompt Optimization},
  author        = {Tao, Wangcheng and Wu, Han and Wong, Weng-Fai},
  year          = {2026},
  eprint        = {2606.04465},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2606.04465}
}


% --- Deployed persistent-update layers (Section 3.2) ---

@misc{google2026geminimemory,
  title        = {{Gemini CLI Auto Memory}},
  author       = {{Google}},
  year         = {2026},
  howpublished = {\url{https://github.com/google-gemini/gemini-cli/blob/main/docs/cli/auto-memory.md}}
}

@misc{github2026copilotmemory,
  title        = {{About GitHub Copilot Memory}},
  author       = {{GitHub}},
  year         = {2026},
  howpublished = {\url{https://docs.github.com/en/copilot/concepts/agents/copilot-memory}}
}

@misc{cursor2026automations,
  title        = {{Automations}},
  author       = {{Cursor}},
  year         = {2026},
  howpublished = {\url{https://cursor.com/changelog/03-05-26}},
  note         = {Cursor Changelog, March 5, 2026}
}

@misc{cognition2026devin,
  title        = {{Devin Session Insights}},
  author       = {{Cognition}},
  year         = {2026},
  howpublished = {\url{https://docs.devin.ai/product-guides/session-insights}}
}

@misc{windsurf2026memories,
  title        = {{Cascade Memories}},
  author       = {{Cognition}},
  year         = {2026},
  howpublished = {\url{https://docs.devin.ai/desktop/cascade/memories}}
}

@misc{replit2026skills,
  title        = {{Replit Agent Skills}},
  author       = {{Replit}},
  year         = {2026},
  howpublished = {\url{https://docs.replit.com/features/agent/skills}}
}

@misc{letta2026code,
  title        = {{Introducing the Letta Code App}},
  author       = {{Letta}},
  year         = {2026},
  howpublished = {\url{https://www.letta.com/blog/introducing-the-letta-code-app/}}
}

@misc{openclaw2026skills,
  title        = {{OpenClaw Skills}},
  author       = {{OpenClaw}},
  year         = {2026},
  howpublished = {\url{https://docs.openclaw.ai/tools/skills}}
}

@misc{augment2025memoryreview,
  title        = {{How We Built Memory Review}},
  author       = {{Augment Code}},
  year         = {2025},
  howpublished = {\url{https://www.augmentcode.com/blog/how-we-built-memory-review}}
}

@misc{primeintellect2026primeagent,
  title        = {{Prime Agent: A Self-Improving RLM Harness}},
  author       = {{Prime Intellect}},
  year         = {2026},
  howpublished = {\url{https://github.com/PrimeIntellect-ai/prime-agent}}
}

@misc{nous2026hermes,
  title        = {{Hermes Agent: Skills System}},
  author       = {{Nous Research}},
  year         = {2026},
  howpublished = {\url{https://hermes-agent.nousresearch.com/docs/user-guide/features/skills}}
}


% Q3 2026 additions verified against arXiv records.
@misc{hu2026evorepair,
  title = {{EvoRepair}: Enhancing Vulnerability Repair Agents Through Experience-Based Self-Evolution},
  author = {Haichuan Hu and Guoqing Xie and Quanjun Zhang and Jiawei Liu and Shengcheng Yu and Chunrong Fang and Zhenyu Chen and Liang Xiao},
  year = {2026},
  date = {2026-05-28},
  eprint = {2605.30105},
  archivePrefix = {arXiv},
  primaryClass = {cs.SE},
  url = {https://arxiv.org/abs/2605.30105}
}

@misc{yue2026ecdysis,
  title = {Ecdysis: Efficient and Effective Training of Runtime Harnesses for {LLM} Agents},
  author = {Ruiqing Yue and Yu Cui and Zhuoyu Sun and Sicheng Pan and Xianhong Xue and Tingyu Li and Ting Li and Wenzhuo Zhu and Yi Chen and Yifei Liu and Baohan Huang and Zhe Cui and Haibin Zhang and Cong Zuo},
  year = {2026},
  date = {2026-09-10},
  eprint = {2609.11677},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.11677}
}

@misc{wu2026modularrsi,
  title = {{ModularRSI}: Modular and Generalizable Recursive Harness Self-Improvement},
  author = {Siwei Wu and Jincheng Ren and Yizhi Li and Haau-Sing Li and Chengran Yang and Xudong Ma and Yuxuan Zhang and Weicheng Gu and Jian Yang and Riza Batista-Navarro and Chuanyi Zhang and Xianglong Liu and Ming Zhou and Bryan Dai and Chenghua Lin},
  year = {2026},
  date = {2026-09-14},
  eprint = {2609.14857},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.14857}
}

@misc{xia2026rrsi,
  title = {{RRSI}: Regularized Recursive Self-Improvement of Agent Harnesses},
  author = {Peng Xia and Rujun Han and Zifeng Wang and Yanfei Chen and Yufan Zhuang and Yoonho Lee and Chengsong Huang and Han Yu and Zhongying CuiZhu and Yifei Ming and Huaxiu Yao and Burak Gokturk and Tomas Pfister and Chen-Yu Lee},
  year = {2026},
  date = {2026-09-21},
  eprint = {2609.24972},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.24972}
}

@misc{du2026livingharness,
  title = {Living-Harness Is an Interactive-Agent Evolver},
  author = {Yuetian Du and Yucheng Wang and He Xu and Jiexu Xu and Shanwen Tan and Bing Zhao and Boyu Yang and Zhijie Xu and Ming Kong and Hu Wei and Jie Liu and Qiang Zhu},
  year = {2026},
  date = {2026-07-29},
  eprint = {2607.26598},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2607.26598}
}

@misc{ursekar2026harnessoptbench,
  title = {{HarnessOpt-Bench}: Evaluating {LLMs} at Harness Optimization},
  author = {Varun Ursekar and Apaar Shanker and Yash Maurya and Shehab Yasser and Vijay S. Kalmath and Veronica Chatrath and Yuan Xue},
  year = {2026},
  date = {2026-08-06},
  eprint = {2608.06301},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2608.06301}
}

@misc{wu2026harnessdev,
  title = {{HarnessDev}: Can {LLMs} Create and Evolve Their Own Agent Harness?},
  author = {Yuhao Wu and Jingyuan Zhang and Jiajun Shi and Xinping Lei and Qingshui Gu and Yuxuan Zhang and Zexuan Wang and Chen He and Chen Huang and Maojia Song and Zhiyuan Zeng and Shaowen Wang and Jinkai Liu and Yunfeng Shi and Jiaheng Liu and Shen Yan and Wenhao Huang and Ge Zhang and Wenxuan Zhang},
  year = {2026},
  date = {2026-09-01},
  eprint = {2609.01437},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.01437}
}

@misc{ke2026evoharnessbench,
  title = {{EvoHarnessBench}: Can Your Agents Keep Pace with an Evolving Harness?},
  author = {Zixuan Ke and Vaidehi Patil and Haizhou Shi and Yang Li and Ye Liu and Sarath Shekkizhar and Anurag Koul and Jiayu Wang and Xuan Phi Nguyen and Semih Yavuz and Mohit Bansal and Shafiq Joty},
  year = {2026},
  date = {2026-09-03},
  eprint = {2609.04280},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.04280}
}

@misc{cai2026reliableharness,
  title = {A Theory of Reliable Self-Evolution for Agent Harnesses},
  author = {Qianshu Cai and Yonggang Zhang and Jun Nie and Maohao Ran and Huajiang Zheng and Jun Song and Xinmei Tian and Yike Guo and Wei Xue},
  year = {2026},
  date = {2026-09-08},
  eprint = {2609.08175},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.08175}
}

@misc{gan2026harnessevolutionlearning,
  title = {Harness Evolution as Learning: Approximation, Generalization, and Optimization Limits of Self-Improving Personal Agents},
  author = {Zeyu Gan and Zixuan Gong and Yong Liu},
  year = {2026},
  date = {2026-09-29},
  eprint = {2609.36892},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.36892}
}

@misc{zhu2026badgenius,
  title = {Bad Genius: Counterfactual-Guided Harness Evolution Beyond Task-Specific Shortcuts},
  author = {Guojun Zhu and Xunheng Huang and Peng Yin and Jiahui Xie and Sanguo Zhang and Doudou Zhou},
  year = {2026},
  date = {2026-09-16},
  eprint = {2609.18366},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.18366}
}

@misc{tian2026ceiling,
  title         = {Harness Evolution Hits a Ceiling: When Weight Training Should Begin},
  author        = {Yuan Tian and Bing Hu and Hao Wang and Binghang Lu and Fang Wu},
  year          = {2026},
  eprint        = {2610.11655},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.11655}
}

@misc{zheng2026servelearnbench,
  title         = {{ServeLearnBench}: How Well Can Agents Self-Improve from Serving Experience?},
  author        = {Haizhong Zheng and Yizhuo Di and Ranajoy Sadhukhan and Shuowei Jin and Beidi Chen},
  year          = {2026},
  eprint        = {2610.07792},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.07792}
}

@misc{wang2026verse,
  title         = {{VERSE}: Verified Self-Evolving Optimizer for Agent Harnesses},
  author        = {Zekai Wang and Yingqiang Ge and Zekun Wang and Hai Wang and Yuhui Xu and Joshua Frandsen and Shancong Fu and Ashia C. Wilson and Chandan K. Reddy},
  year          = {2026},
  eprint        = {2610.02616},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.02616}
}

@misc{yang2026guiharvest,
  title         = {{GUI-HARVEST}: Self-Improving {GUI} Agents through Evidence-Driven Harness Evolution},
  author        = {Geyi Yang and Zikun Qu and Xiang Li and Zhiyong Wang and Min Zhang and Shipei Zeng and Zhongxiang Dai},
  year          = {2026},
  eprint        = {2610.00948},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.00948}
}

@misc{wang2026valve,
  title         = {Certified Long-Horizon Code Agent Evolution via Validation-Gated Skill Optimization},
  author        = {Yifan Wang and Hao Cheng and Xiaomin Li and Yuexing Hao and Hemanth Neelgund Ramesh and Dongwon Jung and Hao Tang and Keru Wang and Chenliang Zhou and Qianhui Wu and Wenlin Yao and Ananth Grama and Andrzej Banburski-Fahey and Baolin Peng and Jaron Lanier and Jianfeng Gao},
  year          = {2026},
  eprint        = {2609.32990},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2609.32990}
}

@misc{singh2026plasticity,
  title         = {Agent Plasticity: Measuring Self-Improvement Through Experience},
  author        = {Harman Singh and Anton Bakhtin and Rulin Shao and Gabriel Synnaeve and Ilia Kulikov and Rob Fergus and Sanjeev Arora and Kurt Keutzer and Jason Weston and Anuj Mahajan and Anirudh Goyal},
  year          = {2026},
  eprint        = {2610.08902},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.08902}
}

@misc{piao2026skillgeneralize,
  title         = {Do Self-Evolving Skills Generalize to Held-Out Tasks?},
  author        = {Xihao Piao and Zifeng Wang and Zhen Chen},
  year          = {2026},
  eprint        = {2609.39148},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2609.39148}
}

@misc{wang2026skillrevisions,
  title         = {Agent Skill Evolution: How Revisions Affect Coding Agents},
  author        = {Jiajie Wang and Yutong Zhao and Tianlin Li and Huashan Chen and Jinfu Chen and Kebin Peng and Sen He},
  year          = {2026},
  eprint        = {2610.04832},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.04832}
}

@misc{das2026seabench,
  title         = {{SEABench}: Benchmarking Endogenous Misalignment in Self-Evolving Agents},
  author        = {Saswat Das and Parvati Viswanathan and Daniel Donnelly and Chang Huang and Sahar Abdelnabi and Ferdinando Fioretto},
  year          = {2026},
  eprint        = {2609.35596},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2609.35596}
}

@misc{li2026rightfit,
  title         = {Finding the Right Fit: Model-Harness Interactions across Agent Tasks},
  author        = {Yixuan Li and Yiyun Zhou and Yao Long Teng and Fuchao Yang and Yanchen Deng and Zhiyi Lyu and Xuyu Dong and Feng Chen and Bo An},
  year          = {2026},
  eprint        = {2610.00917},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.00917}
}

@misc{chen2026cotrace,
  title         = {{CoTrace}: Data Recipes for Training Terminal Agents with Harness-Model Co-Evolution},
  author        = {Jixuan Chen and Jiaxin Zhang and Qinyuan Ye and Yada Pruksachatkun and Haoxiang Zhang and Jingming Zhuo and Yifan Zhang and Yutong Dai and Juntao Tan and Xiangyu Peng and Silvio Savarese and Zeyuan Chen and Lianhui Qin and Chien-Sheng Wu},
  year          = {2026},
  eprint        = {2610.10426},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.10426}
}

@misc{lee2026emho,
  title         = {{EMHO}: Embodied Agent Harness Optimization via Experience Traces},
  author        = {Hyun Jung Lee and Jungtaek Kim and Jongwon Jeong and Tae-Eui Kam and Donghyun Kim and Yong Jae Lee},
  year          = {2026},
  eprint        = {2610.08432},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.08432}
}

% ---- Section 8: education and skill formation (verified at abs/publisher level, 2026-10-10) ----
@article{bastani2025generative,
  title   = {Generative {AI} without guardrails can harm learning: Evidence from high school mathematics},
  author  = {Bastani, Hamsa and Bastani, Osbert and Sungu, Alp and Ge, Haosen and Kabakc{\i}, {\"O}zge and Mariman, Rei},
  journal = {Proceedings of the National Academy of Sciences},
  volume  = {122},
  number  = {26},
  pages   = {e2422633122},
  year    = {2025},
  doi     = {10.1073/pnas.2422633122}
}

@article{kestin2025tutoring,
  title   = {{AI} tutoring outperforms in-class active learning: an {RCT} introducing a novel research-based design in an authentic educational setting},
  author  = {Kestin, Greg and Miller, Kelly and Klales, Anna and Milbourne, Timothy and Ponti, Gregorio},
  journal = {Scientific Reports},
  volume  = {15},
  pages   = {17458},
  year    = {2025},
  doi     = {10.1038/s41598-025-97652-6}
}

@misc{shen2026skill,
  title         = {How {AI} Impacts Skill Formation},
  author        = {Shen, Judy Hanwen and Tamkin, Alex},
  year          = {2026},
  eprint        = {2601.20245},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2601.20245}
}

@misc{lehmann2024classroom,
  title         = {{AI} Meets the Classroom: When Do Large Language Models Harm Learning?},
  author        = {Lehmann, Matthias and Cornelius, Philipp B. and Sting, Fabian J.},
  year          = {2024},
  eprint        = {2409.09047},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2409.09047}
}

@misc{fan2026stock,
  title         = {Toward Measuring {AI}'s Effects on Skill Formation: The Stock-Formation Gap},
  author        = {Fan, Aysa Xuemo},
  year          = {2026},
  eprint        = {2605.16283},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.16283}
}

% ---- Section 8.4: skill and tool-server ecosystem security (verified at arXiv abs level, 2026-10-10) ----
% Venues below are taken from the arXiv comments field.
@misc{liu2026skillsinwild,
  title         = {Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale},
  author        = {Liu, Yi and Wang, Weizhe and Feng, Ruitao and Zhang, Yao and Xu, Guangquan and Deng, Gelei and Li, Yuekang and Zhang, Leo},
  year          = {2026},
  eprint        = {2601.10338},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2601.10338}
}

@inproceedings{liu2026malicious,
  title     = {``{Do} Not Mention This to the User'': Detecting and Understanding Malicious Agent Skills in the Wild},
  author    = {Liu, Yi and Chen, Zhihao and Zhang, Yanjun and Deng, Gelei and Li, Yuekang and Ning, Jianting and Zhang, Leo Yu},
  booktitle = {35th USENIX Security Symposium (USENIX Security 2026)},
  year      = {2026},
  note      = {arXiv:2602.06547}
}

@inproceedings{xiong2026afterparty,
  title     = {After the Party: Growth, Governance, and Security Scanning in the {OpenClaw} Agent Skill Ecosystem},
  author    = {Xiong, Yunpeng and Zhang, Ting},
  booktitle = {33rd Asia-Pacific Software Engineering Conference (APSEC 2026)},
  year      = {2026},
  note      = {To appear; arXiv:2609.17274}
}

@misc{song2025beyond,
  title         = {Beyond the Protocol: Unveiling Attack Vectors in the Model Context Protocol ({MCP}) Ecosystem},
  author        = {Song, Hao and Shen, Yiming and Luo, Wenxuan and Guo, Leixin and Chen, Ting and Wang, Jiashui and Li, Beibei and Zhang, Xiaosong and Chen, Jiachi},
  year          = {2025},
  eprint        = {2506.02040},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2506.02040}
}

@misc{guo2025mcpmeasurement,
  title         = {A Measurement Study of Model Context Protocol Ecosystem},
  author        = {Guo, Hechuan and Hao, Yongle and Zhang, Yue and Xu, Minghui and Lv, Peizhuo and Chen, Jiezhi and Cheng, Xiuzhen},
  year          = {2025},
  eprint        = {2509.25292},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2509.25292}
}

@misc{dong2025minja,
  title         = {Memory Injection Attacks on {LLM} Agents via Query-Only Interaction},
  author        = {Dong, Shen and Xu, Shaochen and He, Pengfei and Li, Yige and Tang, Jiliang and Liu, Tianming and Liu, Hui and Xiang, Zhen},
  year          = {2025},
  eprint        = {2503.03704},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2503.03704}
}

@misc{cohen2024worm,
  title         = {Here Comes The {AI} Worm: Unleashing Zero-click Worms that Target {GenAI}-Powered Applications},
  author        = {Cohen, Stav and Bitton, Ron and Nassi, Ben},
  year          = {2024},
  eprint        = {2403.02817},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2403.02817}
}

% ---- Section 8.4: self-poisoning and monoculture (verified 2026-10-10) ----
@misc{zhang2026skillpoison,
  title         = {{SkillPoison}: Progressive Skill Poisoning via Successful Experiences},
  author        = {Zhang, Lizhi and He, Xin and Fu, Dianxuan and Feng, Yuyuan and Li, Jiatong and Wang, Qi and Wang, Xin and Zhang, Qinggang},
  year          = {2026},
  eprint        = {2610.07645},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2610.07645}
}

@article{kleinberg2021algorithmic,
  title   = {Algorithmic monoculture and social welfare},
  author  = {Kleinberg, Jon and Raghavan, Manish},
  journal = {Proceedings of the National Academy of Sciences},
  volume  = {118},
  number  = {22},
  pages   = {e2018340118},
  year    = {2021},
  doi     = {10.1073/pnas.2018340118}
}

@inproceedings{bommasani2022picking,
  title     = {Picking on the Same Person: Does Algorithmic Monoculture lead to Outcome Homogenization?},
  author    = {Bommasani, Rishi and Creel, Kathleen A. and Kumar, Ananya and Jurafsky, Dan and Liang, Percy},
  booktitle = {Advances in Neural Information Processing Systems},
  volume    = {35},
  year      = {2022}
}

% ---- Section 8.1: research tutors with persistent state (verified 2026-10-10) ----
@misc{zhao2026deeptutor,
  title         = {{DeepTutor}: Towards Agentic Personalized Tutoring},
  author        = {Zhao, Bingxi and Zhang, Jiahao and Ren, Xubin and Guo, Zirui and Chu, Tianzhe and Ma, Yi and Huang, Chao},
  year          = {2026},
  eprint        = {2604.26962},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2604.26962}
}

@misc{wu2025tasa,
  title         = {Teaching According to Students' Aptitude: Personalized Mathematics Tutoring via Persona-, Memory-, and Forgetting-Aware {LLMs}},
  author        = {Wu, Yang and Yao, Rujing and Zhang, Tong and Shi, Yufei and Jiang, Zhuoren and Li, Zhushan and Liu, Xiaozhong},
  year          = {2025},
  eprint        = {2511.15163},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2511.15163}
}

@misc{lee2026llms,
  title         = {{LLMs} Are Already Good Tutors: Training-Free Prompt Optimization for Pedagogical Math Tutoring},
  author        = {Lee, Unggi and Shin, Minchul and Jeong, Yeil and Lee, Sookbun and Moon, Jeongsu and Joo, Kyungtae and Lee, Eunjoo and Kwon, Hoilym},
  year          = {2026},
  eprint        = {2605.27088},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2605.27088}
}


@misc{shen2026dsheverything,
  title        = {{DSH Everything}: {Everything Is a Plugin} and Runtime Self-Evolution},
  author       = {Shen, Yifei},
  year         = {2026},
  howpublished = {Software repository and engineering protocols},
  url          = {https://github.com/yshenaw/dsh-everything/tree/dade643bb09c49d290340ee01f3a00dd20bf37fb},
  note         = {README, EIP self-evolution protocol, and Codex case study; accessed 14 October 2026}
}

% Formal representations and reusable proof structures (Section 9.2).
% Publication metadata checked against Crossref, NeurIPS proceedings,
% and the ICLR camera-ready OpenReview record on 14 October 2026.
@article{beurerkellner2023lmql,
  title   = {Prompting Is Programming: A Query Language for Large Language Models},
  author  = {Beurer-Kellner, Luca and Fischer, Marc and Vechev, Martin},
  journal = {Proceedings of the ACM on Programming Languages},
  volume  = {7},
  number  = {PLDI},
  pages   = {1946--1969},
  year    = {2023},
  doi     = {10.1145/3591300}
}

@inproceedings{demoura2021lean4,
  title     = {The {Lean} 4 Theorem Prover and Programming Language},
  author    = {de Moura, Leonardo and Ullrich, Sebastian},
  booktitle = {Automated Deduction -- CADE 28},
  pages     = {625--635},
  year      = {2021},
  doi       = {10.1007/978-3-030-79876-5_37}
}

@inproceedings{yang2023leandojo,
  title     = {{LeanDojo}: Theorem Proving with Retrieval-Augmented Language Models},
  author    = {Yang, Kaiyu and Swope, Aidan and Gu, Alex and Chalamala, Rahul and Song, Peiyang and Yu, Shixing and Godil, Saad and Prenger, Ryan J. and Anandkumar, Animashree},
  booktitle = {Advances in Neural Information Processing Systems},
  volume    = {36},
  pages     = {21573--21612},
  year      = {2023},
  doi       = {10.52202/075280-0944},
  url       = {https://proceedings.neurips.cc/paper_files/paper/2023/hash/4441469427094f8873d0fecb0c4e1cee-Abstract-Datasets_and_Benchmarks.html}
}

@inproceedings{wang2024legoprover,
  title     = {{LEGO}-Prover: Neural Theorem Proving with Growing Libraries},
  author    = {Wang, Haiming and Xin, Huajian and Zheng, Chuanyang and Liu, Zhengying and Cao, Qingxing and Huang, Yinya and Xiong, Jing and Shi, Han and Xie, Enze and Yin, Jian and Li, Zhenguo and Liang, Xiaodan},
  booktitle = {The Twelfth International Conference on Learning Representations},
  year      = {2024},
  url       = {https://openreview.net/forum?id=3f5PALef5B}
}

% Adaptive practice environments (Section 9.1).
@misc{wang2019poet,
  title         = {Paired Open-Ended Trailblazer ({POET}): Endlessly Generating Increasingly Complex and Diverse Learning Environments and Their Solutions},
  author        = {Wang, Rui and Lehman, Joel and Clune, Jeff and Stanley, Kenneth O.},
  year          = {2019},
  eprint        = {1901.01753},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/1901.01753}
}

@misc{huang2026envharness,
  title         = {{EnvHarness}: Awakening Static Worlds for Agent Learning},
  author        = {Huang, Chengsong and Wang, Zifeng and Han, Rujun and Yan, Jun and Chen, Yanfei and CuiZhu, Zoey and Jiang, Ke and Xia, Peng and Yu, Han and Zhuang, Yufan and Ming, Yifei and Pan, Jiaqi and Mishra, Bhavana Dalvi and Huang, Jiaxin and Gokturk, Burak and Pfister, Tomas and Lee, Chen-Yu},
  year          = {2026},
  eprint        = {2608.19880},
  archivePrefix = {arXiv},
  url           = {https://arxiv.org/abs/2608.19880}
}

@article{wu2026medrsi,
  title = {{MedRSI: Recursive Self-Improvement for Medical Agents via Clinically Aligned Self-Evolution}},
  author = {Wu, Junde and Zhu, Jiayuan and Hu, Minghao and Liu, Fenglin and Pan, Jiazhen},
  journal = {arXiv preprint arXiv:2609.24838},
  year = {2026},
  eprint = {2609.24838},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.24838}
}

@article{xu2026recrsi,
  title = {{Beyond Successor Accuracy: State Retention for Recursive Self-Improvement in Recommendation}},
  author = {Xu, Jinfeng and Chen, Zheyu and Peng, Ziyue and Lin, Zheng and Yuan, Wenhao and Chen, Jian and Li, Shujie and Ngai, Edith},
  journal = {arXiv preprint arXiv:2610.07105},
  year = {2026},
  eprint = {2610.07105},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.07105}
}

@article{li2026jaz,
  title = {{Harness as a Language: A Minimalist Agent Framework With Maximal Expressivity}},
  author = {Li, Zhening and Liu, Joshua and Vukelic, Mateja and Shen, Nicole and Lall, Supriya and Thakur, Amitayush and Zhang, Alex and Khattab, Omar and Light, Jonathan and Solar-Lezama, Armando},
  journal = {arXiv preprint arXiv:2609.26891},
  year = {2026},
  eprint = {2609.26891},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.26891}
}